Publish tested dataset winners and benchmark rankings

.gitignore

@@ -0,0 +1,86 @@
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.egg-info/
+.eggs/
+dist/
+build/
+*.so
+.Python
+.venv/
+venv/
+env/
+.env
+.env.*
+
+# IDE / OS
+.idea/
+.vscode/
+*.swp
+*.swo
+.DS_Store
+Thumbs.db
+
+# ---------------------------------------------------------------------------
+# Upstream change-detection model repositories (clone separately)
+# ---------------------------------------------------------------------------
+BiFA/
+BIT_CD/
+CDMamba/
+Change3D/
+ChangeFormer/
+IFNet/
+RSM-CD/
+SChanger/
+Siam-NestedUNet/
+STANet/
+model_repos/
+
+# Generated dataset layout copies (symlinks/copies of external DATA_ROOT)
+generated_dataset_views/
+
+# ---------------------------------------------------------------------------
+# Model weights, checkpoints, and training artifacts
+# ---------------------------------------------------------------------------
+*.pth
+*.pt
+*.ckpt
+*.safetensors
+*.onnx
+*.pb
+*.h5
+*.pkl
+*.pickle
+*.bin
+*.tar
+*.tar.gz
+*.zip
+
+checkpoints/
+checkpoint/
+experiments/
+exp/
+pretrain/
+pretrained/
+pretrained_weight/
+pretrained_weights/
+weights/
+saved_models/
+saved_model/
+models/*.pth
+models/*.pt
+runs/
+wandb/
+logs/
+vis/
+tmp/
+output/
+outputs/
+samples/
+flagged/
+gradio_cached_examples/
+
+# Large local datasets (point DATA_ROOT outside this repo)
+data/
+datasets/raw/

README.md

@@ -1,3 +1,1571 @@
 ---
-license: mit
+license: other
+library_name: pytorch
+tags:
+  - change-detection
+  - remote-sensing
+  - pytorch
 ---
+
+# CD-Models: A Unified Change Detection Benchmark Suite
+
+> **A multi-dataset, multi-model benchmark for binary remote sensing change detection.**  
+> Covers 21 registered models spanning CNN, Transformer, and Mamba architectures, evaluated across the dataset configs in this repository with a shared training pipeline.
+
+![Python](https://img.shields.io/badge/python-3.10%20recommended-blue)
+![PyTorch](https://img.shields.io/badge/PyTorch-2.6.0%2Bcu124-ee4c2c)
+![Models](https://img.shields.io/badge/models-21-0f766e)
+![Datasets](https://img.shields.io/badge/datasets-8-7c3aed)
+![License](https://img.shields.io/badge/license-see%20upstream-lightgrey)
+
+## Overview
+
+CD-Models is a benchmark harness for binary remote sensing change detection. It brings original research repositories and newly added model wrappers under one launcher, one dataset-config convention, one weight-management layer, and one result aggregation layout. The repository exists to make model comparisons reproducible across the same prepared dataset folders instead of relying on isolated, model-specific scripts. It also keeps the limits visible: result tables are populated only from `results/*/*/metrics_test.json`, and framework-specific models remain marked until their required stacks are installed.
+
+## Supported Models
+
+The model list below is taken from `configs/models/registry.yaml`, which is what `run_training.py` uses for `--model` choices.
+
+| Model | Paper | Venue | Backbone | Params | Training Script | Status |
+|-------|-------|-------|----------|--------|-----------------|--------|
+| `bifa` | BiFA: Remote Sensing Image Change Detection with Bitemporal Feature Alignment | IEEE TGRS 2024 | MixTransformer | — | `train/train_bifa.py` | 🔄 In Progress |
+| `bit_cd` | Remote Sensing Image Change Detection with Transformers | IEEE TGRS 2021 | ResNet-18 + base transformer | ~11M† | `train/train_bit_cd.py` | ✅ Working |
+| `cdmamba` | CDMamba: Incorporating Local Clues into Mamba for Binary Change Detection | IEEE TGRS 2025 | Mamba | — | `train/train_cdmamba.py` | 🔄 In Progress |
+| `cgnet` | Change Guiding Network | IEEE JSTARS 2023 | CGNet custom backbone | — | `train/train_cgnet.py` | ✅ Working |
+| `change3d` | Change3D: Revisiting Change Detection and Captioning from a Video Modeling Perspective | CVPR 2025 Highlight | X3D-L | — | `train/train_change3d.py` | ✅ Working |
+| `changeformer` | A Transformer-Based Siamese Network for Change Detection | IGARSS 2022 | MiT-b4 | ~41M† | `train/train_changeformer.py` | ✅ Working |
+| `changemamba` | ChangeMamba: Remote Sensing Change Detection with Spatio-Temporal State Space Model | IEEE TGRS 2024 | VMamba | — | `train/train_changemamba.py` | ⚠️ Framework Required |
+| `changer` | Changer: Feature Interaction is What You Need for Change Detection | IEEE TGRS / Open-CD | ResNet-18 | — | `train/train_changer.py` | ⚠️ Framework Required |
+| `dsamnet` | Deeply-supervised Attention Metric-based Network | IEEE TGRS 2021 | ResNet | — | `train/train_dsamnet.py` | ✅ Working |
+| `dsifn` | Deeply Supervised Image Fusion Network | ISPRS JPRS 2020 | VGG-16 | — | `train/train_dsifn.py` | 🔄 In Progress |
+| `elgcnet` | ELGC-Net: Efficient Local-Global Context Aggregation | IEEE TGRS 2024 | ResNet-18 / ELGCA | — | `train/train_elgcnet.py` | ✅ Working |
+| `fc_ef` | Fully Convolutional Early Fusion | ICIP 2018 | FCN | — | `train/train_fc_variants.py` | ✅ Working |
+| `fc_siam_conc` | Fully Convolutional Siamese Concatenation | ICIP 2018 | Siamese FCN | — | `train/train_fc_variants.py` | ✅ Working |
+| `fc_siam_diff` | Fully Convolutional Siamese Difference | ICIP 2018 | Siamese FCN | — | `train/train_fc_variants.py` | ✅ Working |
+| `hanet` | HANet: Hierarchical Attention Network | IEEE JSTARS 2023 | HANet custom backbone | — | `train/train_hanet.py` | ✅ Working |
+| `ifnet` | Deeply Supervised Image Fusion Network | ISPRS JPRS 2020 | VGG-16 | — | `train/train_ifnet.py` | 🔄 In Progress |
+| `rsm_cd` | RS-Mamba for Large Remote Sensing Image Dense Prediction | arXiv 2024 | VMamba/RSM-CD tiny | — | `train/train_rsm_cd.py` | ❌ Failed |
+| `schanger` | SChanger: Semantic Change and Spatial Consistency Perspective | IEEE JSTARS 2025 | SChanger-base | — | `train/train_schanger.py` | 🔄 In Progress |
+| `siam_nestedunet` | SNUNet-CD / Siamese NestedUNet | IEEE GRSL 2021 | UNet++ | — | `train/train_siam_nestedunet.py` | 🔄 In Progress |
+| `stanet` | Spatial-Temporal Attention Network | Remote Sensing 2020 | ResNet-18 + PAM | ~17M† | `train/train_stanet.py` | ✅ Working |
+| `tinycd` | TinyCD: A Not So Deep Learning Model for Change Detection | Neural Computing and Applications 2023 | EfficientNet-B4 | — | `train/train_tinycd.py` | ✅ Working |
+
+† Parameter counts are inherited from the previous project README or upstream publications, not measured by the current harness.
+
+## Supported Datasets
+
+Every dataset below is a real YAML file under `configs/datasets/`. The harness expands `${DATA_ROOT}` through `utils/config_loader.py`; if `${DATA_ROOT}` is unset, it first checks `/new-home/buddhiw/Datasets` and then the sibling `mamba-cd/datasets` fallback.
+
+### LEVIR-CD+
+
+| Property | Value |
+|---|---|
+| Task | Binary building change detection |
+| Source | [LEVIR-CD project page](https://justchenhao.github.io/LEVIR/) |
+| Train / Val / Test | Determined by files under `$DATA_ROOT/LEVIR-CD-plus-256/{train,val,test}` |
+| Image size | 256 x 256 |
+| Resolution | Original LEVIR-CD is 0.5 m/pixel; this config uses prepared 256 patches |
+| Channels | RGB |
+| Label convention | Thresholded binary mask; `Mask/` folder, changed pixels above `127` |
+| Mean / Std (A) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Mean / Std (B) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Config | `configs/datasets/levir_cd.yaml` |
+| Root pattern | `${DATA_ROOT}/LEVIR-CD-plus-256` |
+
+### LEVIR-CD+ Test-as-Val
+
+| Property | Value |
+|---|---|
+| Task | Binary building change detection with literature-style test-as-validation protocol |
+| Source | [LEVIR-CD project page](https://justchenhao.github.io/LEVIR/) |
+| Train / Val / Test | Determined by files under `$DATA_ROOT/LEVIR-CD-plus-256-test-as-val/{train,val,test}` |
+| Image size | 256 x 256 |
+| Channels | RGB |
+| Label convention | Thresholded binary mask; `Mask/` folder, changed pixels above `127` |
+| Mean / Std (A) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Mean / Std (B) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Config | `configs/datasets/levir_cd_test_as_val.yaml` |
+| Root pattern | `${DATA_ROOT}/LEVIR-CD-plus-256-test-as-val` |
+
+### WHU-CD
+
+| Property | Value |
+|---|---|
+| Task | Binary building change detection |
+| Source | [WHU building dataset page](https://study.rsgis.whu.edu.cn/pages/download/building_dataset.html) |
+| Train / Val / Test | Determined by files under `$DATA_ROOT/WHU-CD/{train,val,test}` |
+| Image size | 256 x 256 |
+| Channels | RGB |
+| Label convention | Thresholded binary mask; `OUT/` folder, changed pixels above `127` |
+| Mean / Std (A) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Mean / Std (B) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Config | `configs/datasets/whu_cd.yaml` |
+| Root pattern | `${DATA_ROOT}/WHU-CD` |
+
+### DSIFN-CD
+
+| Property | Value |
+|---|---|
+| Task | Binary high-resolution change detection |
+| Source | [DSIFN dataset reference](https://github.com/GeoZcx/A-deeply-supervised-image-fusion-network-for-change-detection-in-remote-sensing-images/tree/master/dataset) |
+| Train / Val / Test | Determined by files under `$DATA_ROOT/DSIFN-CD/DSIFN/{train,val,test}` |
+| Image size | 256 x 256 |
+| Channels | RGB |
+| Label convention | `mask/` folder; raw masks may be `0/1`; generated model views convert to `0/255` where required |
+| Mean / Std (A) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Mean / Std (B) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Config | `configs/datasets/dsifn_cd.yaml` |
+| Root pattern | `${DATA_ROOT}/DSIFN-CD/DSIFN` |
+
+### WildFire-S2
+
+| Property | Value |
+|---|---|
+| Task | Binary burned-area change detection from bi-temporal Sentinel-2 style imagery |
+| Source | Local project dataset card in sibling `WildFire-S2/` |
+| Train / Val / Test | Determined by files under `$DATA_ROOT/WildFireS2/{train,val,test}` |
+| Image size | 256 x 256 |
+| Channels | RGB |
+| Label convention | Thresholded binary mask; `label/` folder, changed pixels above `127` |
+| Mean / Std (A) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Mean / Std (B) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Config | `configs/datasets/wildfire_s2.yaml` |
+| Root pattern | `${DATA_ROOT}/WildFireS2` |
+
+### SYSU-CD
+
+| Property | Value |
+|---|---|
+| Task | Binary change detection |
+| Source | [SYSU-CD project page](https://github.com/liumency/SYSU-CD) |
+| Train / Val / Test | Determined by files under `$DATA_ROOT/SYSU-CD-folders/{train,val,test}` |
+| Image size | 256 x 256 |
+| Channels | RGB |
+| Label convention | Thresholded binary mask; `Mask/` folder, changed pixels above `127` |
+| Mean / Std (A) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Mean / Std (B) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Config | `configs/datasets/sysu_cd.yaml` |
+| Root pattern | `${DATA_ROOT}/SYSU-CD-folders` |
+
+### KATE-CD-256
+
+| Property | Value |
+|---|---|
+| Task | Binary change detection |
+| Source | Local prepared dataset config |
+| Train / Val / Test | Determined by files under `$DATA_ROOT/KATE-CD-256/{train,val,test}` |
+| Image size | 256 x 256 |
+| Channels | RGB |
+| Label convention | Thresholded binary mask; `label/` folder, changed pixels above `127` |
+| Mean / Std (A) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Mean / Std (B) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Config | `configs/datasets/kate_cd.yaml` |
+| Root pattern | `${DATA_ROOT}/KATE-CD-256` |
+
+### Custom-CD
+
+| Property | Value |
+|---|---|
+| Task | User-supplied binary change detection |
+| Source | Local template config |
+| Train / Val / Test | Determined by files under `$DATA_ROOT/Custom-CD/{train,val,test}` |
+| Image size | 256 x 256 |
+| Channels | RGB |
+| Label convention | Thresholded binary mask; `Mask/` folder, changed pixels above `127` |
+| Mean / Std (A) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Mean / Std (B) | `[0.485, 0.456, 0.406]` / `[0.229, 0.224, 0.225]` |
+| Config | `configs/datasets/custom_cd.yaml` |
+| Root pattern | `${DATA_ROOT}/Custom-CD` |
+
+## Results and released checkpoints
+
+Only completed `metrics_test.json` evaluations are eligible. Rankings are computed per dataset by test-set F1 (descending), and only the rank-1 checkpoint for each dataset is released. Validation-only, incomplete, and checkpoint-only runs are excluded.
+
+All accuracy values are fractions. GPU memory is the PyTorch peak reserved memory for inference, with peak allocated memory used only when reserved memory is absent. FPS is model-only throughput. A dash means the evaluator did not record that metric; values are never estimated.
+
+The current evaluator records BF1 but not boundary mean IoU (BmIoU), so BmIoU remains explicitly unavailable rather than being inferred from BF1. Boundary IoU and boundary F1 are distinct measures.
+
+### Released checkpoint summary
+
+| Dataset | Selected model | Test F1 | Download | Size |
+| --- | --- | --- | --- | --- |
+| DSIFN-CD | BIT-CD | 0.6732 | [best_model.pth](https://huggingface.co/dineth18/CD-Models/resolve/main/results/bit_cd/dsifn_cd/checkpoints/best_model.pth?download=true) | 60.8 MiB |
+| LEVIR-CD+ (test-as-val protocol) | CGNet | 0.8138 | [best_model.pth](https://huggingface.co/dineth18/CD-Models/resolve/main/results/cgnet/levir_cd_test_as_val/checkpoints/best_model.pth?download=true) | 405.9 MiB |
+| SYSU-CD | ChangeMamba | 0.8216 | [best_model.pth](https://huggingface.co/dineth18/CD-Models/resolve/main/results/changemamba/sysu_cd/checkpoints/best_model.pth?download=true) | 619.3 MiB |
+| WHU-CD | ChangeMamba | 0.9521 | [best_model.pth](https://huggingface.co/dineth18/CD-Models/resolve/main/results/changemamba/whu_cd/checkpoints/best_model.pth?download=true) | 619.0 MiB |
+
+### DSIFN-CD
+
+| Rank | Model | F1 | mIoU | Overall accuracy | Recall | Precision | BF1 | BmIoU | GFLOPs | GPU (GB) | FPS | Parameters (M) | Checkpoint |
+| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
+| 1 | BIT-CD | 0.6732 | 0.6929 | 0.8919 | 0.6552 | 0.6923 | 0.2819 | — | 10.63 | 2.24 | 127.76 | 12.40 | [Download](https://huggingface.co/dineth18/CD-Models/resolve/main/results/bit_cd/dsifn_cd/checkpoints/best_model.pth?download=true) |
+| 2 | SChanger | 0.6629 | 0.6835 | 0.8857 | 0.6612 | 0.6647 | 0.2847 | — | — | 2.45 | 24.05 | — | — |
+| 3 | ChangeMamba | 0.6435 | 0.6729 | 0.8849 | 0.6113 | 0.6792 | 0.1903 | — | 28.70 | 1.93 | 15.42 | 54.00 | — |
+| 4 | CGNet | 0.6204 | 0.6290 | 0.8343 | 0.7970 | 0.5079 | 0.2903 | — | 82.23 | 6.52 | 65.39 | 38.99 | — |
+| 5 | FC-EF | 0.6078 | 0.6400 | 0.8604 | 0.6365 | 0.5816 | 0.1955 | — | 3.58 | 0.74 | 204.13 | 1.35 | — |
+| 6 | ChangeFormer | 0.5964 | 0.6152 | 0.8299 | 0.7397 | 0.4996 | 0.2422 | — | 21.18 | 4.07 | 49.74 | 29.75 | — |
+| 7 | BiFA | 0.5876 | 0.6111 | 0.8296 | 0.7142 | 0.4991 | 0.2692 | — | 53.00 | 15.10 | 13.81 | 9.87 | — |
+| 8 | Siam-NestedUNet | 0.5839 | 0.6024 | 0.8188 | 0.7482 | 0.4788 | 0.2374 | — | 54.83 | 7.34 | 29.18 | 12.03 | — |
+| 9 | FC-Siam-diff | 0.5829 | 0.6277 | 0.8593 | 0.5784 | 0.5875 | 0.1691 | — | 4.73 | 1.06 | 113.98 | 1.35 | — |
+| 10 | STANet | 0.5483 | 0.6258 | 0.8829 | 0.4184 | 0.7953 | 0.1613 | — | 13.16 | 14.11 | 133.96 | 16.93 | — |
+| 11 | FC-Siam-conc | 0.5479 | 0.5665 | 0.7873 | 0.7583 | 0.4288 | 0.1811 | — | 5.33 | 1.02 | 127.67 | 1.55 | — |
+| 12 | DSAMNet | 0.5467 | 0.5757 | 0.8020 | 0.7027 | 0.4474 | 0.2374 | — | 75.39 | 2.91 | 166.43 | 16.95 | — |
+| 13 | HANet | 0.5347 | 0.5817 | 0.8193 | 0.6109 | 0.4754 | 0.2418 | — | 17.67 | 11.36 | 42.36 | 3.03 | — |
+| 14 | IFNet | 0.4579 | 0.5102 | 0.7524 | 0.6153 | 0.3647 | 0.1325 | — | 82.26 | 6.08 | 104.88 | 50.71 | — |
+
+### LEVIR-CD+ (test-as-val protocol)
+
+| Rank | Model | F1 | mIoU | Overall accuracy | Recall | Precision | BF1 | BmIoU | GFLOPs | GPU (GB) | FPS | Parameters (M) | Checkpoint |
+| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
+| 1 | CGNet | 0.8138 | 0.8356 | 0.9856 | 0.7746 | 0.8572 | 0.7651 | — | 82.23 | 6.01 | 44.01 | 38.99 | [Download](https://huggingface.co/dineth18/CD-Models/resolve/main/results/cgnet/levir_cd_test_as_val/checkpoints/best_model.pth?download=true) |
+| 2 | Siam-NestedUNet | 0.8046 | 0.8282 | 0.9839 | 0.8129 | 0.7965 | 0.7386 | — | — | 2.67 | 108.71 | — | — |
+| 3 | SChanger | 0.8007 | 0.8253 | 0.9836 | 0.8112 | 0.7906 | 0.7364 | — | — | 2.45 | 48.96 | — | — |
+| 4 | BIT-CD | 0.7996 | 0.8247 | 0.9838 | 0.7946 | 0.8047 | 0.7143 | — | 10.63 | 2.27 | 236.60 | 12.40 | — |
+| 5 | HANet | 0.7841 | 0.8131 | 0.9819 | 0.8064 | 0.7630 | 0.7043 | — | — | 5.04 | 84.44 | — | — |
+| 6 | BiFA | 0.7772 | 0.8084 | 0.9818 | 0.7785 | 0.7760 | 0.6581 | — | 53.00 | 15.14 | 30.38 | 9.87 | — |
+| 7 | FC-Siam-conc | 0.7650 | 0.7999 | 0.9809 | 0.7623 | 0.7678 | 0.7175 | — | 5.33 | 1.02 | 308.81 | 1.55 | — |
+| 8 | STANet | 0.7544 | 0.7921 | 0.9791 | 0.7865 | 0.7249 | 0.6388 | — | — | 4.45 | 174.00 | — | — |
+| 9 | ChangeFormer | 0.7488 | 0.7887 | 0.9795 | 0.7505 | 0.7472 | 0.6525 | — | 21.18 | 4.03 | 157.64 | 29.75 | — |
+| 10 | DSAMNet | 0.7444 | 0.7850 | 0.9779 | 0.7897 | 0.7040 | 0.5370 | — | 75.39 | 2.90 | 37.84 | 16.95 | — |
+| 11 | IFNet | 0.7410 | 0.7833 | 0.9786 | 0.7510 | 0.7313 | 0.3489 | — | 82.26 | 6.08 | 105.44 | 50.71 | — |
+| 12 | FC-EF | 0.7319 | 0.7766 | 0.9768 | 0.7785 | 0.6905 | 0.6497 | — | 3.58 | 0.74 | 457.44 | 1.35 | — |
+| 13 | FC-Siam-diff | 0.7196 | 0.7667 | 0.9725 | 0.8650 | 0.6160 | 0.5447 | — | 4.73 | 1.06 | 262.80 | 1.35 | — |
+
+### SYSU-CD
+
+| Rank | Model | F1 | mIoU | Overall accuracy | Recall | Precision | BF1 | BmIoU | GFLOPs | GPU (GB) | FPS | Parameters (M) | Checkpoint |
+| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
+| 1 | ChangeMamba | 0.8216 | 0.8006 | 0.9214 | 0.7677 | 0.8836 | 0.3197 | — | 28.70 | 1.93 | 77.65 | 54.00 | [Download](https://huggingface.co/dineth18/CD-Models/resolve/main/results/changemamba/sysu_cd/checkpoints/best_model.pth?download=true) |
+| 2 | FC-Siam-conc | 0.7778 | 0.7558 | 0.8975 | 0.7609 | 0.7955 | 0.1939 | — | 5.33 | 1.02 | 356.94 | 1.55 | — |
+| 3 | STANet | 0.7728 | 0.7482 | 0.8913 | 0.7841 | 0.7619 | 0.1962 | — | — | 4.45 | 176.02 | — | — |
+| 4 | FC-EF | 0.7710 | 0.7489 | 0.8936 | 0.7596 | 0.7828 | 0.1857 | — | 3.58 | 0.73 | 553.75 | 1.35 | — |
+| 5 | ChangeFormer | 0.7635 | 0.7382 | 0.8851 | 0.7863 | 0.7420 | 0.1591 | — | 21.18 | 4.03 | 143.73 | 29.75 | — |
+| 6 | FC-Siam-diff | 0.7630 | 0.7457 | 0.8956 | 0.7129 | 0.8208 | 0.1853 | — | 4.73 | 1.08 | 552.90 | 1.35 | — |
+| 7 | BIT-CD | 0.7618 | 0.7395 | 0.8881 | 0.7584 | 0.7652 | 0.1780 | — | 10.63 | 2.24 | 111.11 | 12.40 | — |
+| 8 | IFNet | 0.7145 | 0.7020 | 0.8724 | 0.6772 | 0.7562 | 0.1158 | — | — | 3.29 | 109.80 | — | — |
+
+### WHU-CD
+
+| Rank | Model | F1 | mIoU | Overall accuracy | Recall | Precision | BF1 | BmIoU | GFLOPs | GPU (GB) | FPS | Parameters (M) | Checkpoint |
+| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
+| 1 | ChangeMamba | 0.9521 | 0.9519 | 0.9954 | 0.9401 | 0.9644 | 0.8605 | — | 28.70 | 1.93 | 75.28 | 54.00 | [Download](https://huggingface.co/dineth18/CD-Models/resolve/main/results/changemamba/whu_cd/checkpoints/best_model.pth?download=true) |
+| 2 | CGNet | 0.9331 | 0.9340 | 0.9936 | 0.9299 | 0.9364 | 0.8138 | — | 82.23 | 5.62 | 129.54 | 38.99 | — |
+| 3 | BIT-CD | 0.8940 | 0.8989 | 0.9899 | 0.8845 | 0.9037 | 0.6879 | — | 10.63 | 2.24 | 46.28 | 12.40 | — |
+| 4 | DSAMNet | 0.8905 | 0.8956 | 0.9891 | 0.9167 | 0.8657 | 0.5855 | — | 75.39 | 2.90 | 116.58 | 16.95 | — |
+| 5 | IFNet | 0.8389 | 0.8534 | 0.9849 | 0.8147 | 0.8644 | 0.3245 | — | 82.26 | 6.08 | 104.98 | 50.71 | — |
+| 6 | FC-EF | 0.8385 | 0.8531 | 0.9850 | 0.8067 | 0.8729 | 0.5818 | — | 3.58 | 0.74 | 498.55 | 1.35 | — |
+| 7 | FC-Siam-conc | 0.7789 | 0.8062 | 0.9756 | 0.8907 | 0.6920 | 0.4733 | — | 5.33 | 1.03 | 397.86 | 1.55 | — |
+| 8 | FC-Siam-diff | 0.7135 | 0.7580 | 0.9631 | 0.9540 | 0.5699 | 0.2148 | — | 4.73 | 1.09 | 567.87 | 1.35 | — |
+
+### Datasets without a released checkpoint
+
+WildFire-S2, KATE-CD-256, the standard LEVIR-CD+ configuration, and Custom-CD currently have no eligible completed test result. Their validation-only or incomplete checkpoints are intentionally not uploaded.
+
+
+## Installation
+
+### 1. Clone this repository
+
+```bash
+git clone https://github.com/<your-username>/CD-Models.git
+cd CD-Models
+```
+
+### 2. Create and activate environment
+
+```bash
+conda create -n cd-models python=3.10 -y
+conda activate cd-models
+```
+
+### 3. Install PyTorch (CUDA 12.4)
+
+```bash
+pip install torch==2.6.0 torchvision==0.21.0 --index-url https://download.pytorch.org/whl/cu124
+```
+
+Adjust the CUDA version for your system. See [PyTorch Get Started](https://pytorch.org/get-started/locally/).
+
+### 4. Run automatic setup
+
+```bash
+python setup.py
+```
+
+This single command:
+
+- Checks all required Python packages.
+- Auto-installs safe pure-Python/common packages when missing.
+- Clones confirmed model repositories into `model_repos/`.
+- Downloads or warms pretrained backbone weights.
+- Generates dataset list files for datasets already present on disk.
+- Prints a full status report.
+
+### 5. ChangeMamba / Changer MMSeg stack
+
+These two models require OpenMMLab packages that should be installed after PyTorch:
+
+```bash
+pip install mmengine==0.10.1
+pip install mmcv==2.1.0 -f https://download.openmmlab.com/mmcv/dist/cu124/torch2.6/index.html
+pip install mmsegmentation==1.2.2 mmdet==3.3.0 mmpretrain==1.2.0
+```
+
+### 6. Set your dataset root
+
+```bash
+export DATA_ROOT=/path/to/your/datasets
+```
+
+Add the export to `~/.bashrc` or your scheduler job script if you want it to persist.
+
+## Dataset Preparation
+
+All datasets follow the same high-level split structure: `train/`, `val/`, and `test/`, each with a pre-change image folder, a post-change image folder, and a mask folder. Folder names differ by dataset and are encoded in YAML.
+
+### LEVIR-CD+
+
+**Download:** [Official LEVIR page](https://justchenhao.github.io/LEVIR/)
+
+Expected structure:
+
+```text
+$DATA_ROOT/LEVIR-CD-plus-256/
+├── train/
+│   ├── A/
+│   ├── B/
+│   └── Mask/
+├── val/
+│   ├── A/
+│   ├── B/
+│   └── Mask/
+└── test/
+    ├── A/
+    ├── B/
+    └── Mask/
+```
+
+Prepare list files:
+
+```bash
+python setup.py --dataset levir_cd
+```
+
+Config: `configs/datasets/levir_cd.yaml`
+
+### LEVIR-CD+ Test-as-Val
+
+**Download:** Use the same source data as LEVIR-CD+ and prepare the test-as-validation split protocol locally.
+
+Expected structure:
+
+```text
+$DATA_ROOT/LEVIR-CD-plus-256-test-as-val/
+├── train/
+│   ├── A/
+│   ├── B/
+│   └── Mask/
+├── val/
+│   ├── A/
+│   ├── B/
+│   └── Mask/
+└── test/
+    ├── A/
+    ├── B/
+    └── Mask/
+```
+
+Prepare list files:
+
+```bash
+python setup.py --dataset levir_cd_test_as_val
+```
+
+Config: `configs/datasets/levir_cd_test_as_val.yaml`
+
+### WHU-CD
+
+**Download:** [WHU building dataset page](https://study.rsgis.whu.edu.cn/pages/download/building_dataset.html)
+
+Expected structure:
+
+```text
+$DATA_ROOT/WHU-CD/
+├── train/
+│   ├── A/
+│   ├── B/
+│   └── OUT/
+├── val/
+│   ├── A/
+│   ├── B/
+│   └── OUT/
+└── test/
+    ├── A/
+    ├── B/
+    └── OUT/
+```
+
+Prepare list files:
+
+```bash
+python setup.py --dataset whu_cd
+```
+
+Config: `configs/datasets/whu_cd.yaml`
+
+### DSIFN-CD
+
+**Download:** [DSIFN dataset reference](https://github.com/GeoZcx/A-deeply-supervised-image-fusion-network-for-change-detection-in-remote-sensing-images/tree/master/dataset)
+
+Expected structure:
+
+```text
+$DATA_ROOT/DSIFN-CD/DSIFN/
+├── train/
+│   ├── t1/
+│   ├── t2/
+│   └── mask/
+├── val/
+│   ├── t1/
+│   ├── t2/
+│   └── mask/
+└── test/
+    ├── t1/
+    ├── t2/
+    └── mask/
+```
+
+Prepare list files:
+
+```bash
+python setup.py --dataset dsifn_cd
+```
+
+Config: `configs/datasets/dsifn_cd.yaml`
+
+Important DSIFN note: some masks are encoded as `0/1`. The generated compatibility views convert those masks to exact `0/255` PNG data for loaders that use `ToTensor()` and then cast labels to integer classes.
+
+### WildFire-S2
+
+**Download:** Local sibling dataset project, documented in `../WildFire-S2/DATASET_CARD.md`.
+
+Expected structure:
+
+```text
+$DATA_ROOT/WildFireS2/
+├── train/
+│   ├── A/
+│   ├── B/
+│   └── label/
+├── val/
+│   ├── A/
+│   ├── B/
+│   └── label/
+└── test/
+    ├── A/
+    ├── B/
+    └── label/
+```
+
+Prepare list files:
+
+```bash
+python setup.py --dataset wildfire_s2
+```
+
+Config: `configs/datasets/wildfire_s2.yaml`
+
+### SYSU-CD
+
+**Download:** [SYSU-CD repository](https://github.com/liumency/SYSU-CD)
+
+Expected structure:
+
+```text
+$DATA_ROOT/SYSU-CD-folders/
+├── train/
+│   ├── A/
+│   ├── B/
+│   └── Mask/
+├── val/
+│   ├── A/
+│   ├── B/
+│   └── Mask/
+└── test/
+    ├── A/
+    ├── B/
+    └── Mask/
+```
+
+Prepare list files:
+
+```bash
+python setup.py --dataset sysu_cd
+```
+
+Config: `configs/datasets/sysu_cd.yaml`
+
+### KATE-CD-256
+
+**Download:** Local prepared dataset; no official source URL was found in the audited repo files.
+
+Expected structure:
+
+```text
+$DATA_ROOT/KATE-CD-256/
+├── train/
+│   ├── A/
+│   ├── B/
+│   └── label/
+├── val/
+│   ├── A/
+│   ├── B/
+│   └── label/
+└── test/
+    ├── A/
+    ├── B/
+    └── label/
+```
+
+Prepare list files:
+
+```bash
+python setup.py --dataset kate_cd
+```
+
+Config: `configs/datasets/kate_cd.yaml`
+
+### Custom-CD
+
+**Download:** User supplied.
+
+Expected structure:
+
+```text
+$DATA_ROOT/Custom-CD/
+├── train/
+│   ├── A/
+│   ├── B/
+│   └── Mask/
+├── val/
+│   ├── A/
+│   ├── B/
+│   └── Mask/
+└── test/
+    ├── A/
+    ├── B/
+    └── Mask/
+```
+
+Prepare list files:
+
+```bash
+python setup.py --dataset custom_cd
+```
+
+Config: `configs/datasets/custom_cd.yaml`
+
+## Training
+
+All registered models share the same outer interface through `run_training.py`.
+
+### Train a single model on a single dataset
+
+```bash
+python run_training.py --model bifa --dataset levir_cd
+```
+
+### Train a single model on all datasets
+
+```bash
+python run_training.py --model changemamba --dataset all
+```
+
+### Train all models on a single dataset
+
+```bash
+python run_training.py --model all --dataset levir_cd
+```
+
+### Full benchmark sweep
+
+```bash
+python run_training.py --model all --dataset all
+```
+
+Already-completed runs are skipped when `results/{model}/{dataset}/metrics_test.json` exists. To retry anyway:
+
+```bash
+python run_training.py --model bifa --dataset levir_cd --force
+```
+
+Resume interrupted training:
+
+```bash
+python run_training.py --model bifa --dataset levir_cd --resume
+```
+
+Run command-construction checks without training:
+
+```bash
+python run_training.py --model all --dataset dsifn_cd --dry-run
+```
+
+Run a single wrapper smoke test:
+
+```bash
+python train/train_bifa.py --dataset levir_cd --smoke-test
+```
+
+### Available models
+
+Original and legacy integrated models:
+
+```text
+bifa
+bit_cd
+cdmamba
+change3d
+changeformer
+dsifn
+ifnet
+rsm_cd
+schanger
+siam_nestedunet
+stanet
+```
+
+Newly added external or direct adapters:
+
+```text
+fc_ef
+fc_siam_conc
+fc_siam_diff
+changemamba
+elgcnet
+changer
+hanet
+cgnet
+dsamnet
+tinycd
+```
+
+Available datasets:
+
+```text
+custom_cd
+dsifn_cd
+kate_cd
+levir_cd
+levir_cd_test_as_val
+sysu_cd
+whu_cd
+wildfire_s2
+```
+
+### Training outputs
+
+Each completed run is expected to save under:
+
+```text
+results/
+└── bifa/
+    └── levir_cd/
+        ├── checkpoints/
+        │   ├── best_model.pth
+        │   └── latest.pth
+        ├── logs/
+        │   ├── train_stdout.log
+        │   ├── train_stderr.log
+        │   ├── eval_stdout.log
+        │   └── eval_stderr.log
+        ├── predictions/
+        │   ├── test/
+        │   └── test_prob/
+        ├── visuals/
+        │   └── selected_20/
+        ├── metrics_val.json
+        └── metrics_test.json
+```
+
+The master launcher appends run status to `results/training_log.jsonl`. After every run attempt, `utils.results_writer.append_to_comparison_table()` regenerates `results/comparison_table.csv`.
+
+For qualitative comparison, the evaluator writes one deterministic sample manifest per dataset:
+
+```text
+results/qualitative_samples/<dataset>/sample_manifest.json
+```
+
+The same selected test samples are then reused for every model on that dataset. Binary predictions use exact `0` and `255` PNG values.
+
+### Dataset-specific notes
+
+| Model | Dataset | Note |
+|---|---|---|
+| `tinycd` | `wildfire_s2` | Skipped automatically by `run_training.py`; current exclusion is explicit in `DATASET_EXCLUSIONS`. |
+| `changemamba` | all | Requires VMamba weights and the MMSeg/OpenMMLab stack. |
+| `changer` | all | Runs through Open-CD and requires the MMSeg/OpenMMLab stack. |
+| `dsifn` / `ifnet` | `dsifn_cd` | Uses VGG-16 family loaders; DSIFN masks need careful binary scaling. |
+| `rsm_cd` | all | Marked unresolved because the upstream training path has prior initialization/runtime issues. |
+
+## Evaluation
+
+### Evaluate a trained checkpoint
+
+```bash
+python evaluate.py --model bifa --dataset levir_cd
+```
+
+The central evaluator uses `utils/model_adapters.py` for verified in-process PyTorch model construction, checkpoint loading, forward calls, and output normalization. Current smoke-tested adapters are:
+
+```text
+bit_cd
+cgnet
+changeformer
+dsamnet
+elgcnet
+fc_ef
+fc_siam_conc
+fc_siam_diff
+hanet
+siam_nestedunet
+stanet
+tinycd
+```
+
+For these adapters, evaluation loads `checkpoints/best_model.pth`, computes split-level metrics, saves all test predictions, writes selected qualitative panels, records timing/profiling fields, writes `metrics_test.json`, and regenerates the aggregate comparison table.
+
+For remaining subprocess/upstream wrappers, central evaluation fails explicitly until a real architecture construction and checkpoint-loading adapter is added. It does not evaluate dummy models, random initialized models, or guessed output formats.
+
+### Explicit checkpoint path
+
+```bash
+python evaluate.py --model bifa --dataset levir_cd --checkpoint results/bifa/levir_cd/checkpoints/best_model.pth
+```
+
+### Regenerate comparison table
+
+```bash
+python evaluate.py --model all --dataset all --dry-run
+```
+
+### Metrics computed
+
+For binary change detection, metrics are intended for the positive changed class.
+
+| Metric | Description |
+|--------|-------------|
+| F1 | Harmonic mean of precision and recall |
+| IoU | Intersection over Union / Jaccard index |
+| mIoU | Mean IoU over unchanged/background and changed classes |
+| OA | Overall accuracy across both classes |
+| Precision | `TP / (TP + FP)` |
+| Recall | `TP / (TP + FN)` |
+| Kappa | Cohen's kappa coefficient |
+| BF1 | Boundary F1 with configurable tolerance, default 2 pixels |
+| TP / FP / TN / FN | Accumulated confusion counts over the full split |
+
+The implementation lives in `utils/metrics.py`. The evaluator also records measured parameter counts, FLOPs when `thop` or `fvcore` is installed, FPS, PyTorch peak GPU memory, and NVML GPU utilization when `pynvml` is installed. Missing profiler dependencies are recorded with explicit error fields; values are not guessed.
+
+`metrics_test.json` includes the model, dataset, split, checkpoint path, threshold, confusion counts, F1, IoU, mIoU, precision, recall, OA, kappa, BF1, parameter counts, FLOPs fields, FPS fields, GPU profiling fields, test sample count, prediction and visual directories, timestamp, and status.
+
+## Pre-trained Weights
+
+Backbone weights are downloaded or cache-warmed by `python setup.py` and by `utils/weight_downloader.py`.
+
+### Backbone weights
+
+| Model(s) | Backbone | Source | Auto-Downloaded | Size |
+|----------|----------|--------|-----------------|------|
+| ChangeMamba-T | VMamba-Tiny | [Zenodo 14037770](https://zenodo.org/records/14037770) | yes | ~86 MB |
+| ChangeMamba-S | VMamba-Small | [Zenodo 14037770](https://zenodo.org/records/14037770) | yes | ~178 MB |
+| ChangeMamba-B | VMamba-Base | [Zenodo 14037770](https://zenodo.org/records/14037770) | yes | ~391 MB |
+| TinyCD | EfficientNet-B4 | timm | yes | ~75 MB |
+| ChangeFormer, ELGC-Net | MiT variants | timm | yes | 15-82 MB |
+| BIT_CD, STANet, Changer | ResNet-18 | torchvision | yes | ~45 MB |
+| DSAMNet, HANet, CGNet | ResNet-50 / ResNet family | torchvision | yes | ~98 MB |
+| DSIFN / IFNet | VGG-16 | torchvision | yes | ~528 MB |
+
+Pre-download all managed weights:
+
+```bash
+python utils/weight_downloader.py --prefetch-all
+```
+
+List managed VMamba weights:
+
+```bash
+python utils/weight_downloader.py --list
+```
+
+### Trained checkpoints
+
+The released rank-1 test checkpoints and direct download links are listed in [Results and released checkpoints](#results-and-released-checkpoints). Local training still writes checkpoints under `results/{model}/{dataset}/checkpoints/`.
+
+## Repository Structure
+
+```text
+CD-Models/
+├── configs/
+│   ├── datasets/                  # one YAML per dataset
+│   │   ├── custom_cd.yaml
+│   │   ├── dsifn_cd.yaml
+│   │   ├── kate_cd.yaml
+│   │   ├── levir_cd.yaml
+│   │   ├── levir_cd_test_as_val.yaml
+│   │   ├── sysu_cd.yaml
+│   │   ├── whu_cd.yaml
+│   │   └── wildfire_s2.yaml
+│   └── models/                    # model hyperparameters and registry
+│       ├── bifa.yaml
+│       ├── bit_cd.yaml
+│       ├── cdmamba.yaml
+│       ├── cgnet.yaml
+│       ├── change3d.yaml
+│       ├── changeformer.yaml
+│       ├── changemamba.yaml
+│       ├── changer.yaml
+│       ├── dsamnet.yaml
+│       ├── dsifn.yaml
+│       ├── elgcnet.yaml
+│       ├── fc_ef.yaml
+│       ├── fc_siam_conc.yaml
+│       ├── fc_siam_diff.yaml
+│       ├── hanet.yaml
+│       ├── ifnet.yaml
+│       ├── registry.yaml
+│       ├── rsm_cd.yaml
+│       ├── schanger.yaml
+│       ├── siam_nestedunet.yaml
+│       ├── stanet.yaml
+│       └── tinycd.yaml
+├── datasets/
+│   ├── __init__.py
+│   └── cd_dataset.py              # shared dataset validation/loader
+├── train/
+│   ├── fc_adapter.py              # direct FC-EF / FC-Siam trainer
+│   ├── train_bifa.py
+│   ├── train_bit_cd.py
+│   ├── train_cdmamba.py
+│   ├── train_cgnet.py
+│   ├── train_change3d.py
+│   ├── train_changeformer.py
+│   ├── train_changemamba.py
+│   ├── train_changer.py
+│   ├── train_dsamnet.py
+│   ├── train_dsifn.py
+│   ├── train_elgcnet.py
+│   ├── train_fc_variants.py
+│   ├── train_hanet.py
+│   ├── train_ifnet.py
+│   ├── train_rsm_cd.py
+│   ├── train_schanger.py
+│   ├── train_siam_nestedunet.py
+│   ├── train_stanet.py
+│   ├── train_tinycd.py
+│   └── wrapper_common.py           # shared wrapper logic and generated views
+├── utils/
+│   ├── config.py                   # generic YAML/JSON loading and merge helpers
+│   ├── config_loader.py            # dataset/model config expansion and validation
+│   ├── dataset_list_generator.py   # create list/train.txt, list/val.txt, list/test.txt
+│   ├── env_checker.py              # import checks for required packages
+│   ├── legacy_config_writer.py     # generated JSON/list views for legacy trainers
+│   ├── metrics.py                  # binary CD metrics
+│   ├── results.py                  # JSON/CSV helpers
+│   ├── results_writer.py           # metrics_test.json and comparison CSV writer
+│   └── weight_downloader.py        # Zenodo, timm, torchvision weight management
+├── model_repos/                    # cloned external repositories
+│   ├── CGNet-CD/
+│   ├── ChangeMamba/
+│   ├── DSAMNet/
+│   ├── HANet-CD/
+│   ├── TinyCD/
+│   ├── Tiny_model_4_CD/
+│   ├── elgcnet/
+│   ├── fully_convolutional_change_detection/
+│   └── open-cd/
+├── results/
+│   ├── comparison_table.csv
+│   ├── download_log.jsonl
+│   └── training_log.jsonl
+├── generated_configs/              # generated model-specific config files
+├── generated_dataset_views/        # generated compatibility views for upstream repos
+├── BIT_CD/                         # original model repository copy
+├── BiFA/
+├── CDMamba/
+├── Change3D/
+├── ChangeFormer/
+├── IFNet/
+├── RSM-CD/
+├── SChanger/
+├── Siam-NestedUNet/
+├── STANet/
+├── evaluate.py
+├── run_training.py
+├── setup.py
+└── TRAINING_GUIDE.md
+```
+
+## Known Issues
+
+| Model | Issue | Status | Workaround |
+|-------|-------|--------|------------|
+| `rsm_cd` | Upstream RS-Mamba training path has prior initialization/runtime issues. | ❌ Unresolved | Run explicitly only after auditing the upstream environment; skip in broad sweeps if needed. |
+| `changemamba` | Requires VMamba kernels and OpenMMLab stack. | ⚠️ Separate install needed | Install MMSeg stack in Installation section and run `setup.py` for VMamba weights. |
+| `changer` | Requires Open-CD / mmengine / mmcv / mmsegmentation. | ⚠️ Separate install needed | Install MMSeg stack in Installation section. |
+| `tinycd` | RGB-only model path; explicitly excluded for `wildfire_s2`. | ⚠️ By design | Use RGB datasets or edit dataset config only when the data is actually RGB. |
+| `evaluate.py` | Central checkpoint loading is adapter-based and currently smoke-tested for `bit_cd`, `cgnet`, `changeformer`, `dsamnet`, `elgcnet`, FC variants, `hanet`, `siam_nestedunet`, `stanet`, and `tinycd`. Remaining custom/OpenMMLab models still need verified adapters. | 🔄 In Progress | Add a real in-process model construction/checkpoint adapter or call a verified upstream evaluator and convert its outputs. |
+| `dsifn_cd` masks | Raw DSIFN masks may be `0/1`, which some upstream loaders can scale to all zeros. | ✅ Fixed in generated views | Regenerate views by launching training or running dataset preparation. |
+| `SChanger` repo URL | Official clone URL was not confirmed from audited files. | 🔄 In Progress | `setup.py` skips automatic clone for this entry and prints a warning. |
+| root license | No root `LICENSE` file was found during audit. | 🔄 Documentation gap | Check upstream model licenses before redistribution. |
+
+## Adding a New Model
+
+1. Clone the upstream repository into `model_repos/<name>`.
+2. Create `configs/models/<name>.yaml` with `name`, epochs, optimizer, LR, scheduler, loss, backbone, pretrained weights, and image size.
+3. Write `train/train_<name>.py`.
+4. Reuse `train/wrapper_common.py` if the model is a subprocess adapter.
+5. Use `datasets/cd_dataset.py` directly if the model can train in the unified process.
+6. Register the model in `configs/models/registry.yaml`.
+7. Add pretrained weight handling to `utils/weight_downloader.py` or `setup.py` when needed.
+8. Run `python run_training.py --model <name> --dataset levir_cd --dry-run`.
+9. Run the wrapper smoke-test using the real script path, for example `python train/train_bifa.py --dataset levir_cd --smoke-test`.
+10. Run a short real training pass on one dataset before adding the model to a full sweep.
+11. Ensure final metrics are saved as `results/<model>/<dataset>/metrics_test.json`.
+12. Regenerate `results/comparison_table.csv`.
+
+## Citation
+
+If you use this benchmark suite in your research, please cite the individual model papers. Key references:
+
+<details>
+<summary>BibTeX references</summary>
+
+```bibtex
+@article{chen2021bit,
+  title={Remote Sensing Image Change Detection with Transformers},
+  author={Chen, Hao and Qi, Zipeng and Shi, Zhenwei},
+  journal={IEEE Transactions on Geoscience and Remote Sensing},
+  year={2021}
+}
+
+@article{zhang2020dsifn,
+  title={A deeply supervised image fusion network for change detection in high resolution bi-temporal remote sensing images},
+  author={Zhang, Chenxiao and Yue, Peng and Tapete, Deodato and Jiang, Linlin and Shangguan, Bo and Huang, Liangpei and Liu, Guoxiang},
+  journal={ISPRS Journal of Photogrammetry and Remote Sensing},
+  year={2020}
+}
+
+@article{bandara2022changeformer,
+  title={A Transformer-Based Siamese Network for Change Detection},
+  author={Bandara, Wele Gedara Chaminda and Patel, Vishal M.},
+  journal={IGARSS},
+  year={2022}
+}
+
+@article{zhou2025schanger,
+  title={SChanger: Change Detection from a Semantic Change and Spatial Consistency Perspective},
+  author={Zhou, Ziyu and Hu, Keyan and Fang, Yutian and Rui, Xiaoping},
+  journal={IEEE JSTARS},
+  year={2025}
+}
+```
+
+</details>
+
+## Acknowledgements
+
+We thank the authors of all included models for their open-source implementations. This benchmark builds on:
+
+- [Open-CD](https://github.com/likyoo/open-cd) for Changer and the Open-CD toolbox.
+- [VMamba](https://github.com/MzeroMiko/VMamba) for the VMamba backbone family.
+- [timm](https://github.com/huggingface/pytorch-image-models) for pretrained backbone weights.
+- University of Peradeniya for computational resources.
+
+## Audit Appendix: Dataset Config Facts
+
+### Dataset Fact 1
+Config `custom_cd` is defined in `configs/datasets/custom_cd.yaml`.
+Root pattern is `${DATA_ROOT}/Custom-CD`.
+Image size is `256`.
+Channels are `3`.
+Mask folder is `Mask`.
+Pre-change folder is `A`.
+Post-change folder is `B`.
+Splits are `train`, `val`, and `test`.
+Batch size is `8`.
+Workers are `4`.
+Evaluation threshold is `0.5`.
+
+### Dataset Fact 2
+Config `dsifn_cd` is defined in `configs/datasets/dsifn_cd.yaml`.
+Root pattern is `${DATA_ROOT}/DSIFN-CD/DSIFN`.
+Image size is `256`.
+Channels are `3`.
+Mask folder is `mask`.
+Pre-change folder is `t1`.
+Post-change folder is `t2`.
+Splits are `train`, `val`, and `test`.
+Batch size is `8`.
+Workers are `4`.
+Evaluation threshold is `0.5`.
+
+### Dataset Fact 3
+Config `kate_cd` is defined in `configs/datasets/kate_cd.yaml`.
+Root pattern is `${DATA_ROOT}/KATE-CD-256`.
+Image size is `256`.
+Channels are `3`.
+Mask folder is `label`.
+Pre-change folder is `A`.
+Post-change folder is `B`.
+Splits are `train`, `val`, and `test`.
+Batch size is `8`.
+Workers are `4`.
+Evaluation threshold is `0.5`.
+
+### Dataset Fact 4
+Config `levir_cd` is defined in `configs/datasets/levir_cd.yaml`.
+Root pattern is `${DATA_ROOT}/LEVIR-CD-plus-256`.
+Image size is `256`.
+Channels are `3`.
+Mask folder is `Mask`.
+Pre-change folder is `A`.
+Post-change folder is `B`.
+Splits are `train`, `val`, and `test`.
+Batch size is `8`.
+Workers are `4`.
+Evaluation threshold is `0.5`.
+
+### Dataset Fact 5
+Config `levir_cd_test_as_val` is defined in `configs/datasets/levir_cd_test_as_val.yaml`.
+Root pattern is `${DATA_ROOT}/LEVIR-CD-plus-256-test-as-val`.
+Image size is `256`.
+Channels are `3`.
+Mask folder is `Mask`.
+Pre-change folder is `A`.
+Post-change folder is `B`.
+Splits are `train`, `val`, and `test`.
+Protocol is `test_as_val_literature`.
+Batch size is `8`.
+Workers are `4`.
+
+### Dataset Fact 6
+Config `sysu_cd` is defined in `configs/datasets/sysu_cd.yaml`.
+Root pattern is `${DATA_ROOT}/SYSU-CD-folders`.
+Image size is `256`.
+Channels are `3`.
+Mask folder is `Mask`.
+Pre-change folder is `A`.
+Post-change folder is `B`.
+Splits are `train`, `val`, and `test`.
+Batch size is `8`.
+Workers are `4`.
+Evaluation threshold is `0.5`.
+
+### Dataset Fact 7
+Config `whu_cd` is defined in `configs/datasets/whu_cd.yaml`.
+Root pattern is `${DATA_ROOT}/WHU-CD`.
+Image size is `256`.
+Channels are `3`.
+Mask folder is `OUT`.
+Pre-change folder is `A`.
+Post-change folder is `B`.
+Splits are `train`, `val`, and `test`.
+Batch size is `8`.
+Workers are `4`.
+Evaluation threshold is `0.5`.
+
+### Dataset Fact 8
+Config `wildfire_s2` is defined in `configs/datasets/wildfire_s2.yaml`.
+Root pattern is `${DATA_ROOT}/WildFireS2`.
+Image size is `256`.
+Channels are `3`.
+Mask folder is `label`.
+Pre-change folder is `A`.
+Post-change folder is `B`.
+Splits are `train`, `val`, and `test`.
+Batch size is `8`.
+Workers are `4`.
+Evaluation threshold is `0.5`.
+
+## Audit Appendix: Model Config Facts
+
+### Model Fact 1
+Model `bifa` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/bifa.yaml`.
+Training script is `train/train_bifa.py`.
+Source path is `BiFA`.
+Framework is `pytorch`.
+Backbone is `MixTransformer`.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `ce_dice`.
+
+### Model Fact 2
+Model `bit_cd` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/bit_cd.yaml`.
+Training script is `train/train_bit_cd.py`.
+Source path is `BIT_CD`.
+Framework is `pytorch`.
+Backbone is `ResNet-18 + base_transformer_pos_s4_dd8`.
+Optimizer is `sgd`.
+Learning rate is `0.01`.
+Epochs are `200`.
+Loss is `cross_entropy`.
+
+### Model Fact 3
+Model `cdmamba` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/cdmamba.yaml`.
+Training script is `train/train_cdmamba.py`.
+Source path is `CDMamba`.
+Framework is `pytorch`.
+Backbone is `Mamba`.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `ce_dice`.
+
+### Model Fact 4
+Model `cgnet` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/cgnet.yaml`.
+Training script is `train/train_cgnet.py`.
+Source path is `model_repos/CGNet-CD`.
+Framework is `pytorch`.
+Backbone is `null` in YAML and implemented by upstream CGNet.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `original_cgnet_loss`.
+
+### Model Fact 5
+Model `change3d` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/change3d.yaml`.
+Training script is `train/train_change3d.py`.
+Source path is `Change3D`.
+Framework is `pytorch`.
+Backbone is `X3D-L`.
+Optimizer is `adam`.
+Learning rate is `0.0002`.
+Epochs are `200`.
+Loss is `bce_dice`.
+
+### Model Fact 6
+Model `changeformer` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/changeformer.yaml`.
+Training script is `train/train_changeformer.py`.
+Source path is `ChangeFormer`.
+Framework is `pytorch`.
+Backbone is `MiT-b4`.
+Optimizer is `adamw`.
+Learning rate is `0.00006`.
+Epochs are `200`.
+Loss is `cross_entropy`.
+
+### Model Fact 7
+Model `changemamba` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/changemamba.yaml`.
+Training script is `train/train_changemamba.py`.
+Source path is `model_repos/ChangeMamba`.
+Framework is `mmsegmentation`.
+Backbone is `VMamba`.
+Optimizer is `adamw`.
+Learning rate is `0.00006`.
+Epochs are `200`.
+Loss is `mmseg_decode_head_loss`.
+
+### Model Fact 8
+Model `changer` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/changer.yaml`.
+Training script is `train/train_changer.py`.
+Source path is `model_repos/open-cd`.
+Framework is `mmsegmentation`.
+Backbone is `ResNet-18`.
+Optimizer is `adamw`.
+Learning rate is `0.00006`.
+Epochs are `200`.
+Loss is `mmseg_decode_head_loss`.
+
+### Model Fact 9
+Model `dsamnet` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/dsamnet.yaml`.
+Training script is `train/train_dsamnet.py`.
+Source path is `model_repos/DSAMNet`.
+Framework is `pytorch`.
+Backbone is `ResNet`.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `contrastive_triplet_bce_original`.
+
+### Model Fact 10
+Model `dsifn` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/dsifn.yaml`.
+Training script is `train/train_dsifn.py`.
+Source path is `IFNet`.
+Framework is `pytorch`.
+Backbone is `VGG-16`.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `dsifn_original`.
+
+### Model Fact 11
+Model `elgcnet` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/elgcnet.yaml`.
+Training script is `train/train_elgcnet.py`.
+Source path is `model_repos/elgcnet`.
+Framework is `pytorch`.
+Backbone is `ResNet-18`.
+Optimizer is `sgd`.
+Learning rate is `0.01`.
+Epochs are `200`.
+Loss is `cross_entropy`.
+
+### Model Fact 12
+Model `fc_ef` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/fc_ef.yaml`.
+Training script is `train/train_fc_variants.py`.
+Source path is `model_repos/fully_convolutional_change_detection`.
+Framework is `pytorch`.
+Backbone is `null`.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `bce_dice`.
+
+### Model Fact 13
+Model `fc_siam_conc` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/fc_siam_conc.yaml`.
+Training script is `train/train_fc_variants.py`.
+Source path is `model_repos/fully_convolutional_change_detection`.
+Framework is `pytorch`.
+Backbone is `null`.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `bce_dice`.
+
+### Model Fact 14
+Model `fc_siam_diff` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/fc_siam_diff.yaml`.
+Training script is `train/train_fc_variants.py`.
+Source path is `model_repos/fully_convolutional_change_detection`.
+Framework is `pytorch`.
+Backbone is `null`.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `bce_dice`.
+
+### Model Fact 15
+Model `hanet` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/hanet.yaml`.
+Training script is `train/train_hanet.py`.
+Source path is `model_repos/HANet-CD`.
+Framework is `pytorch`.
+Backbone is `null` in YAML and implemented by upstream HANet.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `original_hanet_loss`.
+
+### Model Fact 16
+Model `ifnet` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/ifnet.yaml`.
+Training script is `train/train_ifnet.py`.
+Source path is `IFNet`.
+Framework is `pytorch`.
+Backbone is `VGG-16`.
+Optimizer is `adam`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `dsifn_cd_loss`.
+
+### Model Fact 17
+Model `rsm_cd` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/rsm_cd.yaml`.
+Training script is `train/train_rsm_cd.py`.
+Source path is `RSM-CD/change_detection_mamba`.
+Framework is `pytorch`.
+Backbone is `VMamba/RSM-CD tiny`.
+Optimizer is `adamw`.
+Learning rate is `0.001`.
+Epochs are `200`.
+Loss is `fccdn_loss_without_seg`.
+
+### Model Fact 18
+Model `schanger` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/schanger.yaml`.
+Training script is `train/train_schanger.py`.
+Source path is `SChanger`.
+Framework is `pytorch`.
+Backbone is `SChanger-base`.
+Optimizer is `adamw`.
+Learning rate is `0.00006`.
+Epochs are `200`.
+Loss is `bce_deep_supervision`.
+
+### Model Fact 19
+Model `siam_nestedunet` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/siam_nestedunet.yaml`.
+Training script is `train/train_siam_nestedunet.py`.
+Source path is `Siam-NestedUNet`.
+Framework is `pytorch`.
+Backbone is `UNet++`.
+Optimizer is `adam`.
+Learning rate is `0.00015`.
+Epochs are `200`.
+Loss is `cross_entropy`.
+
+### Model Fact 20
+Model `stanet` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/stanet.yaml`.
+Training script is `train/train_stanet.py`.
+Source path is `STANet`.
+Framework is `pytorch`.
+Backbone is `ResNet-18 + PAM`.
+Optimizer is `adam`.
+Learning rate is `0.001`.
+Epochs are `200`.
+Loss is `bcl_contrastive`.
+
+### Model Fact 21
+Model `tinycd` is registered in `configs/models/registry.yaml`.
+Config path is `configs/models/tinycd.yaml`.
+Training script is `train/train_tinycd.py`.
+Source path is `model_repos/Tiny_model_4_CD`.
+Framework is `pytorch`.
+Backbone is `EfficientNet-B4`.
+Optimizer is `adamw`.
+Learning rate is `0.0001`.
+Epochs are `200`.
+Loss is `bce_dice`.
+
+## Audit Appendix: Utility Roles
+
+### Utility Fact 1
+`utils/config.py` provides generic YAML and JSON loading.
+It also provides `deep_merge`.
+It exposes a simple model registry loader.
+It does not expand `${DATA_ROOT}`.
+
+### Utility Fact 2
+`utils/config_loader.py` provides the current dataset and model config API.
+It expands `${DATA_ROOT}`.
+It checks `/new-home/buddhiw/Datasets` by default.
+It checks the sibling `mamba-cd/datasets` fallback.
+It attaches `_config_path` and `_dataset_name`.
+
+### Utility Fact 3
+`utils/dataset_list_generator.py` writes `list/train.txt`, `list/val.txt`, and `list/test.txt`.
+It scans image folders such as `A`, `T1`, `t1`, `img`, `images`, or `image`.
+It is idempotent unless `overwrite=True`.
+
+### Utility Fact 4
+`utils/env_checker.py` checks imports for torch, torchvision, numpy, Pillow, OpenCV, PyYAML, tqdm, timm, einops, mmengine, mmcv, mmseg, and gdown.
+It prints install hints.
+It does not auto-install packages.
+
+### Utility Fact 5
+`utils/legacy_config_writer.py` prepares list-based generated views for legacy trainers.
+It writes generated JSON configs for BiFA and CDMamba.
+It converts `0/1` masks to lossless `0/255` PNG data where needed.
+
+### Utility Fact 6
+`utils/metrics.py` implements binary metrics.
+It supports class-logit tensors.
+It supports thresholded probability maps.
+It returns F1, IoU, OA, precision, recall, kappa, and confusion counts.
+
+### Utility Fact 7
+`utils/results.py` writes JSON and append-only CSV-like result files.
+It adds a UTC timestamp when appending a result row.
+
+### Utility Fact 8
+`utils/results_writer.py` writes `metrics_<split>.json`.
+It scans `results/*/*/metrics_test.json`.
+It regenerates `results/comparison_table.csv`.
+
+### Utility Fact 9
+`utils/weight_downloader.py` manages VMamba Zenodo/GDrive downloads.
+It warms timm models.
+It warms torchvision models.
+It logs downloads to `results/download_log.jsonl`.
+
+## Audit Appendix: Wrapper Modes
+
+### Wrapper Fact 1
+`bifa` uses the shared wrapper and launches `BiFA/train_cd.py` for non-default datasets.
+It writes generated JSON through `utils/legacy_config_writer.py`.
+It is subprocess mode.
+
+### Wrapper Fact 2
+`bit_cd` uses the shared wrapper and launches `BIT_CD/main_cd.py`.
+It uses generated legacy list views for non-default datasets.
+It is subprocess mode.
+
+### Wrapper Fact 3
+`cdmamba` uses the shared wrapper and launches `CDMamba/train_cd.py`.
+It writes generated JSON through `utils/legacy_config_writer.py`.
+It is subprocess mode.
+
+### Wrapper Fact 4
+`changeformer` uses the shared wrapper and launches `ChangeFormer/main_cd.py`.
+It uses generated legacy list views.
+It is subprocess mode.
+
+### Wrapper Fact 5
+`dsamnet` launches `model_repos/DSAMNet/train.py`.
+It passes direct train and validation folder paths.
+It warms torchvision ResNet weights before training.
+It is subprocess mode.
+
+### Wrapper Fact 6
+`cgnet` launches `model_repos/CGNet-CD/train_CGNet.py`.
+It builds matched split views under `generated_dataset_views/`.
+It warms torchvision ResNet-50 weights before training.
+It is subprocess mode.
+
+### Wrapper Fact 7
+`hanet` launches `model_repos/HANet-CD/trainHCX.py`.
+It writes a generated metadata JSON.
+It builds matched split views under `generated_dataset_views/`.
+It is subprocess mode.
+
+### Wrapper Fact 8
+`tinycd` launches `model_repos/Tiny_model_4_CD/training.py`.
+It builds TinyCD list views.
+It warms timm EfficientNet-B4.
+It is subprocess mode.
+
+### Wrapper Fact 9
+`changer` launches `model_repos/open-cd/tools/train.py`.
+It writes an Open-CD config under `generated_configs/`.
+It requires mmengine, mmcv, and mmsegmentation.
+It is subprocess mode.
+
+### Wrapper Fact 10
+`changemamba` launches `model_repos/ChangeMamba/changedetection/script/train_MambaBCD.py`.
+It builds ChangeMamba `T1`, `T2`, and `GT` views.
+It uses VMamba-Tiny weights from Zenodo record `14037770`.
+It is subprocess mode.
+
+### Wrapper Fact 11
+`fc_ef`, `fc_siam_conc`, and `fc_siam_diff` use `train/fc_adapter.py`.
+They train directly in this repository process.
+They use `datasets/cd_dataset.py`.
+They save validation metrics to `results/<model>/<dataset>/metrics_val.json`.
+
+### Wrapper Fact 12
+`run_training.py` skips missing dataset roots before launching wrappers.
+It skips completed model-dataset pairs when `metrics_test.json` exists.
+It supports `--dry-run`, `--resume`, `--eval-only`, `--gpu`, and `--force`.
+
+### Wrapper Fact 13
+`evaluate.py` can resolve datasets and checkpoints.
+It regenerates the comparison CSV.
+It still needs architecture-specific model loaders.
+
+## Audit Appendix: Clone Status
+
+### Cloned External Repositories
+`model_repos/CGNet-CD` exists.
+`model_repos/ChangeMamba` exists.
+`model_repos/DSAMNet` exists.
+`model_repos/HANet-CD` exists.
+`model_repos/TinyCD` exists.
+`model_repos/Tiny_model_4_CD` exists.
+`model_repos/elgcnet` exists.
+`model_repos/fully_convolutional_change_detection` exists.
+`model_repos/open-cd` exists.
+
+### Root-Level Model Repositories
+`BIT_CD` exists.
+`BiFA` exists.
+`CDMamba` exists.
+`Change3D` exists.
+`ChangeFormer` exists.
+`IFNet` exists.
+`RSM-CD` exists.
+`SChanger` exists.
+`Siam-NestedUNet` exists.
+`STANet` exists.
+
+### Missing or Unconfirmed Clone Targets
+`model_repos/DSIFN` is not present.
+The official DSIFN URL for that exact clone target was not confirmed from audited files.
+`setup.py` therefore does not guess it.
+`SChanger` has a root-level copy, but the setup script marks its clone URL as unconfirmed.
+
+## Audit Appendix: Command Reference
+
+### Setup Commands
+`python setup.py`
+Runs environment checks, clones repos, fetches weights, and prepares dataset lists.
+`python setup.py --status`
+Prints status without changes.
+`python setup.py --skip-weights`
+Skips weight downloads.
+`python setup.py --env-check-only`
+Checks packages only.
+`python setup.py --dataset levir_cd`
+Prepares list files for one dataset config.
+
+### Training Commands
+`python run_training.py --model bifa --dataset levir_cd`
+Trains one model on one dataset.
+`python run_training.py --model all --dataset levir_cd`
+Runs every registered model on one dataset.
+`python run_training.py --model bifa --dataset all`
+Runs one model across all configured datasets with available roots.
+`python run_training.py --model all --dataset all`
+Runs the full configured sweep.
+`python run_training.py --model all --dataset dsifn_cd --dry-run`
+Checks launch commands without training.
+
+### Evaluation Commands
+`python evaluate.py --model bifa --dataset levir_cd`
+Attempts central evaluation for a trained checkpoint.
+`python evaluate.py --model all --dataset all --dry-run`
+Checks evaluation config resolution.
+`python utils/weight_downloader.py --prefetch-all`
+Fetches managed pretrained weights.
+`python utils/weight_downloader.py --list`
+Lists managed VMamba weight files.
+
+## Audit Appendix: Result Provenance
+
+### Current Result State
+No `metrics_test.json` files were found during the audit.
+`results/comparison_table.csv` contains only a header row.
+Fresh benchmark numbers must therefore be produced by new training and evaluation runs.
+
+### Historical Result State
+The old README contained WildFire-S2 checkpoint numbers for `change3d`, `bit_cd`, `changeformer`, and `stanet`.
+Those numbers remain in this README only with a dagger marker.
+They are not presented as current unified-harness results.
+
+### Reporting Rule
+A number in the Results section must come from either `metrics_test.json` or a daggered historical note.
+Unknown values are shown as `—`.
+Empty cells are avoided.
+
+## Audit Appendix: Environment Snapshot
+
+The local `Mamba` environment used during audit reported:
+
+```text
+torch 2.6.0+cu124
+torchvision 0.21.0+cu124
+numpy 2.1.2
+PIL 12.2.0
+cv2 4.13.0
+yaml 6.0.2
+tqdm 4.68.1
+timm 0.4.12
+einops 0.8.1
+sklearn 1.6.1
+```
+
+`timm 0.4.12` does not expose every modern MiT alias. The weight downloader treats missing `mit_b0` and `mit_b1` aliases as optional warmup warnings unless the model explicitly requires them.

TRAINING_GUIDE.md

@@ -0,0 +1,160 @@
+# CD-Models Training Guide
+
+## Prerequisites
+
+- Python 3.10 is recommended.
+- CUDA/PyTorch requirements vary by model. Check each model directory or cloned repo requirements file before training:
+  - `BiFA/requirements.txt`
+  - `CDMamba/requirement.txt`
+  - `Change3D/requirements.txt`
+  - `ChangeFormer/requirements.txt`
+  - `RSM-CD/requirements.txt`
+- Set `DATA_ROOT` before using central configs:
+
+```bash
+export DATA_ROOT=/path/to/datasets
+```
+
+All central configs use `${DATA_ROOT}` and avoid hardcoded absolute dataset paths.
+
+## Dataset Setup
+
+The shared loader expects split folders under each dataset root. Folder names are configured in `configs/datasets/*.yaml`.
+
+LEVIR-CD+:
+
+```text
+$DATA_ROOT/LEVIR-CD-plus-256/
+  train/A train/B train/Mask
+  val/A   val/B   val/Mask
+  test/A  test/B  test/Mask
+```
+
+WHU-CD:
+
+```text
+$DATA_ROOT/WHU-CD/
+  train/A train/B train/OUT
+  val/A   val/B   val/OUT
+  test/A  test/B  test/OUT
+```
+
+WildFireS2:
+
+```text
+$DATA_ROOT/WildFireS2/
+  train/A train/B train/label
+  val/A   val/B   val/label
+  test/A  test/B  test/label
+```
+
+Download sources should be confirmed from the original dataset pages before publication. Keep raw data outside this repo and point `DATA_ROOT` at the prepared benchmark folders.
+
+## Pretrained Weights
+
+| Model | Backbone | Download URL | Local Path |
+|---|---|---|---|
+| change3d | X3D-L | Original Change3D README/model release | `Change3D/model/X3D_L.pyth` |
+| ifnet / dsifn | VGG-16 | Torchvision or original DSIFN README | `model_repos/DSIFN/pretrained/vgg16.pth` |
+| changemamba | VMamba | ChangeMamba README Zenodo/HuggingFace link | `model_repos/ChangeMamba/pretrained/vmamba_tiny.pth` |
+| changer | ResNet-18 | open-cd / torchvision ResNet-18 | `model_repos/open-cd/pretrained/resnet18.pth` |
+| tinycd | EfficientNet-B4 | TinyCD README | `model_repos/TinyCD/pretrained/efficientnet_b4.pth` |
+
+Wrappers print a clear required-weight block when a required file is missing.
+
+## Automatic Weight Downloads
+
+All pretrained backbone weights are downloaded automatically on first training run. VMamba weights for ChangeMamba are downloaded from Zenodo record `14037770` by `utils/weight_downloader.py`.
+
+| Model | Backbone | Download Source | Auto? | Est. Size |
+|---|---|---|---|---|
+| ChangeMamba | VMamba-Tiny | Zenodo records/14037770 | yes | ~86 MB |
+| ChangeMamba | VMamba-Small | Zenodo records/14037770 | yes | ~178 MB |
+| ChangeMamba | VMamba-Base | Zenodo records/14037770 | yes | ~391 MB |
+| Changer | ResNet-18 | PyTorch Hub / torchvision | yes | ~45 MB |
+| DSAMNet | ResNet-18 | PyTorch Hub / torchvision | yes | ~45 MB |
+| TinyCD | EfficientNet-B4 | timm / torchvision backend | yes | ~75 MB |
+| HANet | ResNet-50 | PyTorch Hub / torchvision | yes | ~98 MB |
+| CGNet | ResNet-50 | PyTorch Hub / torchvision | yes | ~98 MB |
+| DSIFN | VGG-16 | PyTorch Hub / torchvision | yes | ~528 MB |
+| ELGC-Net | MiT-b0 | timm / HuggingFace | yes | ~15 MB |
+| ChangeFormer | MiT-b1 | timm / HuggingFace | yes | ~28 MB |
+
+To pre-download all managed weights before training:
+
+```bash
+python utils/weight_downloader.py --prefetch-all
+```
+
+## Quick Start
+
+Train one model on one dataset:
+
+```bash
+python run_training.py --model bifa --dataset wildfire_s2
+```
+
+Train all models on one dataset:
+
+```bash
+python run_training.py --model all --dataset wildfire_s2 --dry-run
+```
+
+Run evaluation only:
+
+```bash
+python run_training.py --model bifa --dataset wildfire_s2 --eval-only
+```
+
+## Adding A New Model
+
+1. Confirm the official GitHub URL and license.
+2. Clone into `model_repos/<model_name>`.
+3. Read the cloned README, configs, model code, and training code.
+4. Add `configs/models/<model_name>.yaml`.
+5. Add or update `train/train_<model_name>.py`.
+6. Register the model in `configs/models/registry.yaml`.
+7. Ensure the wrapper accepts `--dataset`, `--gpu`, `--resume`, `--eval-only`, and `--dry-run`.
+8. Save final metrics with `utils.results_writer.save_metrics()`.
+
+## Results
+
+Per-run metrics are written to:
+
+```text
+results/<model>/<dataset>/metrics_test.json
+```
+
+The aggregate table is regenerated at:
+
+```text
+results/comparison_table.csv
+```
+
+Columns: `Model`, `Dataset`, `F1`, `IoU`, `OA`, `Precision`, `Recall`, `Kappa`.
+
+## Known Issues
+
+### Existing WildFire Wrappers
+
+**Partially resolved.** Central wrappers now accept every configured dataset and validate the requested dataset through `CDDataset`. The original nested research scripts still contain their own model-specific data loops, so full non-WildFire training requires each upstream loop to be run against compatible prepared data or rewritten in detail.
+
+### ChangeMamba
+
+**Resolved for setup.** Cloned to `model_repos/ChangeMamba`. Requires running `python model_repos/ChangeMamba/install_deps.py` once before training. VMamba pretrained weights are downloaded automatically from Zenodo record `14037770` on first training run.
+
+### Changer
+
+**Resolved for setup.** Uses open-cd cloned to `model_repos/open-cd`. It requires the same mmengine/mmcv family of dependencies as open-cd. ResNet-18 weights download automatically through torchvision on first run.
+
+### DSAMNet
+
+**Resolved for setup.** Cloned from `liumency/DSAMNet`. The original repo exposes `loss/BCL.py` and `loss/DiceLoss.py`; wrappers must use that metric-learning loss path rather than replacing it with BCE+Dice. Labels are expected as binary 0/1 maps; `CDDataset` converts 255-valued masks to 1.
+
+### TinyCD
+
+**Resolved for setup.** Cloned from `AndreaCodegoni/Tiny_model_4_CD`. EfficientNet-B4 weights download automatically through the weight downloader/timm warmup. TinyCD is RGB-only; `run_training.py` skips `tinycd/wildfire_s2` by default.
+
+### HANet / CGNet
+
+**Resolved for setup.** Cloned from `ChengxiHAN/HANet-CD` and `ChengxiHAN/CGNet-CD`. Both repos use similar custom training utilities and include dataset helpers under `utils/`. ResNet-50 weights are warmed automatically through torchvision.

configs/datasets/custom_cd.yaml

@@ -0,0 +1,29 @@
+name: custom_cd
+data_root: ${DATA_ROOT}/Custom-CD
+source_name: Custom-CD
+img_size: 256
+num_classes: 2
+ignore_index: 255
+channels: 3
+mean_a: [0.485, 0.456, 0.406]
+std_a: [0.229, 0.224, 0.225]
+mean_b: [0.485, 0.456, 0.406]
+std_b: [0.229, 0.224, 0.225]
+mask_threshold: 127
+mask_folder: Mask
+image_a_folder: A
+image_b_folder: B
+splits:
+  train: train
+  val: val
+  test: test
+batch_size: 8
+num_workers: 4
+eval:
+  threshold: 0.5
+  sweep_val_threshold: true
+  threshold_min: 0.05
+  threshold_max: 0.95
+  threshold_step: 0.05
+  boundary_tolerance: 2
+

configs/datasets/dsifn_cd.yaml

@@ -0,0 +1,28 @@
+name: dsifn_cd
+data_root: ${DATA_ROOT}/DSIFN-CD/DSIFN
+source_name: DSIFN-CD
+img_size: 256
+num_classes: 2
+ignore_index: 255
+channels: 3
+mean_a: [0.485, 0.456, 0.406]
+std_a: [0.229, 0.224, 0.225]
+mean_b: [0.485, 0.456, 0.406]
+std_b: [0.229, 0.224, 0.225]
+mask_threshold: 127
+mask_folder: mask
+image_a_folder: t1
+image_b_folder: t2
+splits:
+  train: train
+  val: val
+  test: test
+batch_size: 8
+num_workers: 4
+eval:
+  threshold: 0.5
+  sweep_val_threshold: true
+  threshold_min: 0.05
+  threshold_max: 0.95
+  threshold_step: 0.05
+  boundary_tolerance: 2

configs/datasets/kate_cd.yaml

@@ -0,0 +1,29 @@
+name: kate_cd
+data_root: ${DATA_ROOT}/KATE-CD-256
+source_name: KATE-CD-256
+img_size: 256
+num_classes: 2
+ignore_index: 255
+channels: 3
+mean_a: [0.485, 0.456, 0.406]
+std_a: [0.229, 0.224, 0.225]
+mean_b: [0.485, 0.456, 0.406]
+std_b: [0.229, 0.224, 0.225]
+mask_threshold: 127
+mask_folder: label
+image_a_folder: A
+image_b_folder: B
+splits:
+  train: train
+  val: val
+  test: test
+batch_size: 8
+num_workers: 4
+eval:
+  threshold: 0.5
+  sweep_val_threshold: true
+  threshold_min: 0.05
+  threshold_max: 0.95
+  threshold_step: 0.05
+  boundary_tolerance: 2
+

configs/datasets/levir_cd.yaml

@@ -0,0 +1,29 @@
+name: levir_cd
+data_root: ${DATA_ROOT}/LEVIR-CD-plus-256-test-as-val
+source_name: LEVIR-CD+
+img_size: 256
+num_classes: 2
+ignore_index: 255
+channels: 3
+mean_a: [0.485, 0.456, 0.406]
+std_a: [0.229, 0.224, 0.225]
+mean_b: [0.485, 0.456, 0.406]
+std_b: [0.229, 0.224, 0.225]
+mask_threshold: 127
+mask_folder: Mask
+image_a_folder: A
+image_b_folder: B
+splits:
+  train: train
+  val: val
+  test: test
+batch_size: 8
+num_workers: 4
+eval:
+  threshold: 0.5
+  sweep_val_threshold: true
+  threshold_min: 0.4
+  threshold_max: 0.95
+  threshold_step: 0.02
+  boundary_tolerance: 2
+

configs/datasets/levir_cd_test_as_val.yaml

@@ -0,0 +1,26 @@
+name: levir_cd_test_as_val
+data_root: ${DATA_ROOT}/LEVIR-CD-plus-256-test-as-val
+source_name: LEVIR-CD+
+img_size: 256
+num_classes: 2
+ignore_index: 255
+channels: 3
+mean_a: [0.485, 0.456, 0.406]
+std_a: [0.229, 0.224, 0.225]
+mean_b: [0.485, 0.456, 0.406]
+std_b: [0.229, 0.224, 0.225]
+mask_threshold: 127
+mask_folder: Mask
+image_a_folder: A
+image_b_folder: B
+splits:
+  train: train
+  val: val
+  test: test
+batch_size: 8
+num_workers: 4
+protocol: test_as_val_literature
+eval:
+  threshold: 0.5
+  sweep_val_threshold: false
+  boundary_tolerance: 2

configs/datasets/sysu_cd.yaml

@@ -0,0 +1,28 @@
+name: sysu_cd
+data_root: ${DATA_ROOT}/SYSU-CD-folders
+source_name: SYSU-CD
+img_size: 256
+num_classes: 2
+ignore_index: 255
+channels: 3
+mean_a: [0.485, 0.456, 0.406]
+std_a: [0.229, 0.224, 0.225]
+mean_b: [0.485, 0.456, 0.406]
+std_b: [0.229, 0.224, 0.225]
+mask_threshold: 127
+mask_folder: Mask
+image_a_folder: A
+image_b_folder: B
+splits:
+  train: train
+  val: val
+  test: test
+batch_size: 8
+num_workers: 4
+eval:
+  threshold: 0.5
+  sweep_val_threshold: true
+  threshold_min: 0.05
+  threshold_max: 0.95
+  threshold_step: 0.05
+  boundary_tolerance: 2

configs/datasets/whu_cd.yaml

@@ -0,0 +1,29 @@
+name: whu_cd
+data_root: ${DATA_ROOT}/WHU-CD
+source_name: WHU-CD
+img_size: 256
+num_classes: 2
+ignore_index: 255
+channels: 3
+mean_a: [0.485, 0.456, 0.406]
+std_a: [0.229, 0.224, 0.225]
+mean_b: [0.485, 0.456, 0.406]
+std_b: [0.229, 0.224, 0.225]
+mask_threshold: 127
+mask_folder: OUT
+image_a_folder: A
+image_b_folder: B
+splits:
+  train: train
+  val: val
+  test: test
+batch_size: 8
+num_workers: 4
+eval:
+  threshold: 0.5
+  sweep_val_threshold: true
+  threshold_min: 0.05
+  threshold_max: 0.95
+  threshold_step: 0.05
+  boundary_tolerance: 2
+

configs/datasets/wildfire_s2.yaml

@@ -0,0 +1,29 @@
+name: wildfire_s2
+data_root: ${DATA_ROOT}/WildFireS2
+source_name: WildFireS2
+img_size: 256
+num_classes: 2
+ignore_index: 255
+channels: 3
+mean_a: [0.485, 0.456, 0.406]
+std_a: [0.229, 0.224, 0.225]
+mean_b: [0.485, 0.456, 0.406]
+std_b: [0.229, 0.224, 0.225]
+mask_threshold: 127
+mask_folder: label
+image_a_folder: A
+image_b_folder: B
+splits:
+  train: train
+  val: val
+  test: test
+batch_size: 8
+num_workers: 4
+eval:
+  threshold: 0.5
+  sweep_val_threshold: true
+  threshold_min: 0.05
+  threshold_max: 0.95
+  threshold_step: 0.05
+  boundary_tolerance: 2
+

configs/models/bifa.yaml

@@ -0,0 +1,11 @@
+name: bifa
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0
+scheduler: linear
+loss: ce_dice
+backbone: MixTransformer
+pretrained_weights: null
+img_size: 256
+

configs/models/bit_cd.yaml

@@ -0,0 +1,11 @@
+name: bit_cd
+num_epochs: 200
+optimizer: sgd
+lr: 0.01
+weight_decay: 0.0005
+scheduler: cosine
+loss: cross_entropy
+backbone: ResNet-18 + base_transformer_pos_s4_dd8
+pretrained_weights: null
+img_size: 256
+

configs/models/cdmamba.yaml

@@ -0,0 +1,11 @@
+name: cdmamba
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0
+scheduler: linear
+loss: ce_dice
+backbone: Mamba
+pretrained_weights: null
+img_size: 256
+

configs/models/cgnet.yaml

@@ -0,0 +1,11 @@
+name: cgnet
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0001
+scheduler: cosine
+loss: original_cgnet_loss
+backbone: null
+pretrained_weights: null
+img_size: 256
+

configs/models/change3d.yaml

@@ -0,0 +1,11 @@
+name: change3d
+num_epochs: 200
+optimizer: adam
+lr: 0.0002
+weight_decay: 0.0001
+scheduler: poly
+loss: bce_dice
+backbone: X3D-L
+pretrained_weights: Change3D/model/X3D_L.pyth
+img_size: 256
+

configs/models/changeformer.yaml

@@ -0,0 +1,11 @@
+name: changeformer
+num_epochs: 200
+optimizer: adamw
+lr: 0.00006
+weight_decay: 0.01
+scheduler: cosine
+loss: cross_entropy
+backbone: MiT-b4
+pretrained_weights: null
+img_size: 256
+

configs/models/changemamba.yaml

@@ -0,0 +1,14 @@
+name: changemamba
+num_epochs: 200
+max_iters: 50000
+smoke_max_iters: 2
+optimizer: adamw
+lr: 0.00006
+weight_decay: 0.01
+scheduler: poly
+loss: mmseg_decode_head_loss
+backbone: VMamba-Small
+vmamba_variant: small
+changemamba_model_type: MambaBCD_Small
+pretrained_weights: model_repos/ChangeMamba/pretrained_weight/vssmsmall_dp03_ckpt_epoch_238.pth
+img_size: 256

configs/models/changer.yaml

@@ -0,0 +1,11 @@
+name: changer
+num_epochs: 200
+optimizer: adamw
+lr: 0.00006
+weight_decay: 0.01
+scheduler: poly
+loss: mmseg_decode_head_loss
+backbone: ResNet-18
+pretrained_weights: null
+img_size: 256
+

configs/models/dsamnet.yaml

@@ -0,0 +1,11 @@
+name: dsamnet
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0001
+scheduler: step
+loss: contrastive_triplet_bce_original
+backbone: ResNet
+pretrained_weights: null
+img_size: 256
+

configs/models/dsifn.yaml

@@ -0,0 +1,11 @@
+name: dsifn
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0001
+scheduler: step
+loss: dsifn_original
+backbone: VGG-16
+pretrained_weights: model_repos/DSIFN/pretrained/vgg16.pth
+img_size: 256
+

configs/models/elgcnet.yaml

@@ -0,0 +1,11 @@
+name: elgcnet
+num_epochs: 200
+optimizer: sgd
+lr: 0.01
+weight_decay: 0.0005
+scheduler: cosine
+loss: cross_entropy
+backbone: ResNet-18
+pretrained_weights: null
+img_size: 256
+

configs/models/fc_ef.yaml

@@ -0,0 +1,11 @@
+name: fc_ef
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0
+scheduler: step
+loss: bce_dice
+backbone: null
+pretrained_weights: null
+img_size: 256
+

configs/models/fc_siam_conc.yaml

@@ -0,0 +1,11 @@
+name: fc_siam_conc
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0
+scheduler: step
+loss: bce_dice
+backbone: null
+pretrained_weights: null
+img_size: 256
+

configs/models/fc_siam_diff.yaml

@@ -0,0 +1,11 @@
+name: fc_siam_diff
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0
+scheduler: step
+loss: bce_dice
+backbone: null
+pretrained_weights: null
+img_size: 256
+

configs/models/hanet.yaml

@@ -0,0 +1,11 @@
+name: hanet
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0001
+scheduler: cosine
+loss: original_hanet_loss
+backbone: null
+pretrained_weights: null
+img_size: 256
+

configs/models/ifnet.yaml

@@ -0,0 +1,11 @@
+name: ifnet
+num_epochs: 200
+optimizer: adam
+lr: 0.0001
+weight_decay: 0.0001
+scheduler: step
+loss: dsifn_cd_loss
+backbone: VGG-16
+pretrained_weights: torchvision://vgg16
+img_size: 256
+

configs/models/registry.yaml

@@ -0,0 +1,106 @@
+models:
+  bifa:
+    script: train/train_bifa.py
+    source_path: BiFA
+    framework: pytorch
+    pretrained_backbone: MixTransformer
+  bit_cd:
+    script: train/train_bit_cd.py
+    source_path: BIT_CD
+    framework: pytorch
+    pretrained_backbone: ResNet-18
+  cdmamba:
+    script: train/train_cdmamba.py
+    source_path: CDMamba
+    framework: pytorch
+    pretrained_backbone: Mamba
+  change3d:
+    script: train/train_change3d.py
+    source_path: Change3D
+    framework: pytorch
+    pretrained_backbone: X3D-L
+  changeformer:
+    script: train/train_changeformer.py
+    source_path: ChangeFormer
+    framework: pytorch
+    pretrained_backbone: MiT-b4
+  ifnet:
+    script: train/train_ifnet.py
+    source_path: IFNet
+    framework: pytorch
+    pretrained_backbone: VGG-16
+  rsm_cd:
+    script: train/train_rsm_cd.py
+    source_path: RSM-CD/change_detection_mamba
+    framework: pytorch
+    pretrained_backbone: VMamba
+  schanger:
+    script: train/train_schanger.py
+    source_path: SChanger
+    framework: pytorch
+    pretrained_backbone: null
+  siam_nestedunet:
+    script: train/train_siam_nestedunet.py
+    source_path: Siam-NestedUNet
+    framework: pytorch
+    pretrained_backbone: null
+  stanet:
+    script: train/train_stanet.py
+    source_path: STANet
+    framework: pytorch
+    pretrained_backbone: ResNet-18
+  fc_ef:
+    script: train/train_fc_variants.py
+    source_path: model_repos/fully_convolutional_change_detection
+    framework: pytorch
+    pretrained_backbone: null
+  fc_siam_conc:
+    script: train/train_fc_variants.py
+    source_path: model_repos/fully_convolutional_change_detection
+    framework: pytorch
+    pretrained_backbone: null
+  fc_siam_diff:
+    script: train/train_fc_variants.py
+    source_path: model_repos/fully_convolutional_change_detection
+    framework: pytorch
+    pretrained_backbone: null
+  dsifn:
+    script: train/train_dsifn.py
+    source_path: IFNet
+    framework: pytorch
+    pretrained_backbone: VGG-16
+  changemamba:
+    script: train/train_changemamba.py
+    source_path: model_repos/ChangeMamba
+    framework: mmsegmentation
+    pretrained_backbone: VMamba
+  elgcnet:
+    script: train/train_elgcnet.py
+    source_path: model_repos/elgcnet
+    framework: pytorch
+    pretrained_backbone: ResNet-18
+  changer:
+    script: train/train_changer.py
+    source_path: model_repos/open-cd
+    framework: mmsegmentation
+    pretrained_backbone: ResNet-18
+  hanet:
+    script: train/train_hanet.py
+    source_path: model_repos/HANet-CD
+    framework: pytorch
+    pretrained_backbone: null
+  cgnet:
+    script: train/train_cgnet.py
+    source_path: model_repos/CGNet-CD
+    framework: pytorch
+    pretrained_backbone: null
+  dsamnet:
+    script: train/train_dsamnet.py
+    source_path: model_repos/DSAMNet
+    framework: pytorch
+    pretrained_backbone: ResNet
+  tinycd:
+    script: train/train_tinycd.py
+    source_path: model_repos/Tiny_model_4_CD
+    framework: pytorch
+    pretrained_backbone: EfficientNet-B4

configs/models/rsm_cd.yaml

@@ -0,0 +1,11 @@
+name: rsm_cd
+num_epochs: 200
+optimizer: adamw
+lr: 0.001
+weight_decay: 0.001
+scheduler: reduce_on_plateau
+loss: fccdn_loss_without_seg
+backbone: VMamba/RSM-CD tiny
+pretrained_weights: null
+img_size: 256
+

configs/models/schanger.yaml

@@ -0,0 +1,11 @@
+name: schanger
+num_epochs: 200
+optimizer: adamw
+lr: 0.00006
+weight_decay: 0.0001
+scheduler: cosine
+loss: bce_deep_supervision
+backbone: SChanger-base
+pretrained_weights: null
+img_size: 256
+

configs/models/siam_nestedunet.yaml

@@ -0,0 +1,11 @@
+name: siam_nestedunet
+num_epochs: 200
+optimizer: adam
+lr: 0.00015
+weight_decay: 0.0001
+scheduler: step
+loss: cross_entropy
+backbone: UNet++
+pretrained_weights: null
+img_size: 256
+

configs/models/stanet.yaml

@@ -0,0 +1,11 @@
+name: stanet
+num_epochs: 200
+optimizer: adam
+lr: 0.001
+weight_decay: 0.0
+scheduler: cosine
+loss: bcl_contrastive
+backbone: ResNet-18 + PAM
+pretrained_weights: null
+img_size: 256
+

configs/models/tinycd.yaml

@@ -0,0 +1,11 @@
+name: tinycd
+num_epochs: 200
+optimizer: adamw
+lr: 0.0001
+weight_decay: 0.01
+scheduler: cosine
+loss: bce_dice
+backbone: EfficientNet-B4
+pretrained_weights: model_repos/TinyCD/pretrained/efficientnet_b4.pth
+img_size: 256
+

datasets/__init__.py

@@ -0,0 +1,4 @@
+from .cd_dataset import CDDataset, build_dataloader
+
+__all__ = ["CDDataset", "build_dataloader"]
+

datasets/cd_dataset.py

@@ -0,0 +1,198 @@
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Dict, Iterable, Optional
+
+import numpy as np
+import torch
+from PIL import Image
+from torch.utils.data import DataLoader, Dataset
+from utils.dataset_cache import dataloader_kwargs, print_dataloader_policy
+
+try:
+    from torchvision import transforms
+except Exception:  # pragma: no cover - torchvision import errors are environment-specific.
+    transforms = None
+
+
+IMG_EXTS = {".png", ".jpg", ".jpeg", ".tif", ".tiff", ".bmp"}
+MASK_DIR_CANDIDATES = ("Mask", "mask", "label", "labels", "OUT", "gt")
+
+
+def _scan(folder: Path) -> Dict[str, Path]:
+    return {
+        p.stem: p
+        for p in sorted(folder.iterdir())
+        if p.is_file() and not p.name.startswith(".") and p.suffix.lower() in IMG_EXTS
+    }
+
+
+def _resolve(root: Path, value: str | Path) -> Path:
+    p = Path(value)
+    return p if p.is_absolute() else root / p
+
+
+def _configured_dirs(root: Path, split: str, cfg: dict) -> Optional[tuple[Path, Path, Path]]:
+    if "splits" in cfg:
+        split_dir = root / cfg.get("splits", {}).get(split, split)
+        a_dir = split_dir / cfg.get("image_a_folder", "A")
+        b_dir = split_dir / cfg.get("image_b_folder", "B")
+        m_dir = split_dir / cfg.get("mask_folder", "label")
+        return a_dir, b_dir, m_dir
+    source = cfg.get("split", {}).get("source_folders", {})
+    a_rel = source.get(f"{split}_a") or source.get("a")
+    b_rel = source.get(f"{split}_b") or source.get("b")
+    m_rel = source.get(f"{split}_mask") or source.get("mask")
+    if not (a_rel and b_rel and m_rel):
+        return None
+    return _resolve(root, a_rel), _resolve(root, b_rel), _resolve(root, m_rel)
+
+
+def _mask_dir(split_dir: Path) -> Path:
+    for name in MASK_DIR_CANDIDATES:
+        candidate = split_dir / name
+        if candidate.is_dir():
+            return candidate
+    tried = ", ".join(MASK_DIR_CANDIDATES)
+    raise FileNotFoundError(f"No mask directory under {split_dir}; tried {tried}")
+
+
+class CDDataset(Dataset):
+    """Configurable binary change-detection dataset.
+
+    This is the shared root loader for CD-Models. It intentionally supports the
+    folder conventions found in mamba-cd and the nested research repos instead
+    of adding another WildFire-specific loader.
+    """
+
+    def __init__(
+        self,
+        root: str | Path,
+        split: str,
+        cfg: Optional[dict] = None,
+        image_size: Optional[int] = None,
+        normalize: bool = True,
+        return_format: str = "dict",
+    ) -> None:
+        self.root = Path(root)
+        self.split = split
+        self.cfg = cfg or {}
+        self.return_format = return_format
+        ds_cfg = self.cfg.get("dataset", self.cfg)
+        self.image_size = int(image_size or ds_cfg.get("image_size", ds_cfg.get("img_size", 256)))
+        self.threshold = int(ds_cfg.get("binary_threshold", ds_cfg.get("mask_threshold", 127)))
+        self.mean = ds_cfg.get("mean", ds_cfg.get("mean_a", [0.485, 0.456, 0.406]))
+        self.std = ds_cfg.get("std", ds_cfg.get("std_a", [0.229, 0.224, 0.225]))
+        if not self.root.is_dir():
+            dataset_name = ds_cfg.get("_dataset_name", ds_cfg.get("name", "unknown"))
+            raise FileNotFoundError(
+                f"\n{'=' * 60}\n"
+                f"Dataset root directory not found:\n  {self.root}\n\n"
+                f"For dataset '{dataset_name}', either:\n"
+                f"  1. Download the dataset and place it at the above path\n"
+                f"  2. Update DATA_ROOT: export DATA_ROOT=/correct/path\n"
+                f"  3. Edit configs/datasets/{dataset_name}.yaml\n"
+                f"{'=' * 60}"
+            )
+
+        configured = _configured_dirs(self.root, split, self.cfg)
+        if configured is None:
+            split_dir = self.root / split
+            a_dir = split_dir / "A"
+            b_dir = split_dir / "B"
+            m_dir = _mask_dir(split_dir)
+        else:
+            a_dir, b_dir, m_dir = configured
+
+        for label, folder in (("A", a_dir), ("B", b_dir), ("mask", m_dir)):
+            if not folder.is_dir():
+                raise FileNotFoundError(f"{label} directory not found: {folder}")
+
+        a_files = _scan(a_dir)
+        b_files = _scan(b_dir)
+        m_files = _scan(m_dir)
+        stems = sorted(set(a_files) & set(b_files) & set(m_files))
+        if not stems:
+            raise RuntimeError(f"No matched A/B/mask triplets found for {self.root} split {split}")
+        self.samples = [(a_files[s], b_files[s], m_files[s], s) for s in stems]
+
+        if transforms is None:
+            self.image_tf = None
+        else:
+            ops: list = [transforms.Resize((self.image_size, self.image_size)), transforms.ToTensor()]
+            if normalize:
+                ops.append(transforms.Normalize(mean=self.mean, std=self.std))
+            self.image_tf = transforms.Compose(ops)
+
+    def __len__(self) -> int:
+        return len(self.samples)
+
+    def _image(self, path: Path) -> torch.Tensor:
+        img = Image.open(path).convert("RGB")
+        if self.image_tf is not None:
+            return self.image_tf(img)
+        img = img.resize((self.image_size, self.image_size), Image.BILINEAR)
+        arr = np.array(img, dtype=np.float32) / 255.0
+        return torch.from_numpy(arr).permute(2, 0, 1)
+
+    def _mask(self, path: Path) -> torch.Tensor:
+        mask = Image.open(path).convert("L").resize((self.image_size, self.image_size), Image.NEAREST)
+        arr = np.array(mask, dtype=np.uint8)
+        if arr.max() <= 1:
+            bin_mask = (arr > 0).astype(np.float32)
+        else:
+            bin_mask = (arr > self.threshold).astype(np.float32)
+        return torch.from_numpy(bin_mask).unsqueeze(0)
+
+    def __getitem__(self, index: int):
+        a_path, b_path, mask_path, name = self.samples[index]
+        a = self._image(a_path)
+        b = self._image(b_path)
+        mask = self._mask(mask_path)
+        if self.return_format == "tuple":
+            return a, b, mask, name
+        if self.return_format == "legacy":
+            return {"A": a, "B": b, "L": mask.squeeze(0).long(), "name": name}
+        return {"a": a, "b": b, "mask": mask, "name": name}
+
+
+def build_dataloader(
+    cfg: dict,
+    split: str,
+    shuffle: Optional[bool] = None,
+    return_format: str = "dict",
+) -> DataLoader:
+    ds_cfg = cfg.get("dataset", cfg)
+    dataset = CDDataset(
+        ds_cfg.get("root", ds_cfg["data_root"]),
+        split=split,
+        cfg=cfg,
+        image_size=int(ds_cfg.get("image_size", ds_cfg.get("img_size", 256))),
+        return_format=return_format,
+    )
+    batch_size = int(ds_cfg.get("batch_size", 8))
+    num_workers = int(ds_cfg.get("num_workers", 4))
+    if shuffle is None:
+        shuffle = split == "train"
+    print_dataloader_policy({**ds_cfg, "num_workers": num_workers}, torch.cuda.is_available())
+    return DataLoader(
+        dataset,
+        batch_size=batch_size,
+        shuffle=shuffle,
+        **dataloader_kwargs({**ds_cfg, "num_workers": num_workers}, torch.cuda.is_available()),
+        drop_last=split == "train",
+    )
+
+
+def available_splits(cfg: dict) -> Iterable[str]:
+    ds_cfg = cfg.get("dataset", cfg)
+    root = Path(ds_cfg.get("root", ds_cfg["data_root"]))
+    if "splits" in cfg:
+        yield from cfg["splits"]
+        return
+    source = cfg.get("split", {}).get("source_folders", {})
+    for split in ("train", "val", "test"):
+        if source.get(f"{split}_a") and source.get(f"{split}_b") and source.get(f"{split}_mask"):
+            yield split
+        elif (root / split).is_dir():
+            yield split

evaluate.py

@@ -0,0 +1,142 @@
+"""
+evaluate.py - Run test sweep on a trained cd-model checkpoint.
+
+Usage:
+    python evaluate.py --model changemamba --dataset levir_cd
+    python evaluate.py --model all --dataset all
+    python evaluate.py --checkpoint results/bifa/levir_cd/checkpoints/best_model.pth --model bifa --dataset levir_cd
+"""
+from __future__ import annotations
+
+import argparse
+import sys
+from pathlib import Path
+
+import torch
+import yaml
+
+from utils.config_loader import load_dataset_config, load_model_config
+from utils.gpu_utils import print_gpu_diagnostics, resolve_gpu
+from utils.model_adapters import get_model_adapter
+from utils.results_writer import append_to_comparison_table
+from utils.unified_evaluator import evaluate_with_adapter
+
+
+ROOT = Path(__file__).resolve().parent
+if str(ROOT) not in sys.path:
+    sys.path.insert(0, str(ROOT))
+
+class ModelEvaluationUnavailable(RuntimeError):
+    pass
+
+
+def _load_datasets() -> list[str]:
+    return sorted(path.stem for path in (ROOT / "configs" / "datasets").glob("*.yaml"))
+
+
+def _load_registry() -> dict:
+    with (ROOT / "configs" / "models" / "registry.yaml").open("r", encoding="utf-8") as f:
+        return yaml.safe_load(f)["models"]
+
+
+def _best_checkpoint(model: str, dataset: str) -> Path:
+    ckpt_dir = ROOT / "results" / model / dataset / "checkpoints"
+    direct = ckpt_dir / "best_model.pth"
+    if direct.exists():
+        return direct
+    candidates = (
+        sorted(ckpt_dir.glob("*F1*.pth"))
+        + sorted(ckpt_dir.glob("*.pth"))
+        + sorted(ckpt_dir.glob("**/best_ckpt.pth"))
+        + sorted(ckpt_dir.glob("**/best_ckpt.pt"))
+        + sorted(ckpt_dir.glob("**/*best*.pth"))
+        + sorted(ckpt_dir.glob("**/*best*.pt"))
+    )
+    if not candidates:
+        raise FileNotFoundError(f"No checkpoint found under {ckpt_dir}")
+    return candidates[-1]
+
+
+def _parse_args() -> argparse.Namespace:
+    registry = _load_registry()
+    datasets = _load_datasets()
+    parser = argparse.ArgumentParser(description="Run a test sweep on a trained cd-model checkpoint.")
+    parser.add_argument("--model", required=True, choices=sorted(registry) + ["all"])
+    parser.add_argument("--dataset", required=True, choices=datasets + ["all"])
+    parser.add_argument("--checkpoint", default=None)
+    parser.add_argument("--gpu", default="0")
+    parser.add_argument("--dry-run", action="store_true")
+    parser.add_argument("--allow-missing-profilers", action="store_true")
+    return parser.parse_args()
+
+
+def _evaluate_one(
+    model: str,
+    dataset: str,
+    checkpoint: str | None,
+    gpu: str,
+    dry_run: bool,
+    allow_missing_profilers: bool,
+) -> int:
+    cfg = load_dataset_config(dataset)
+    gpu_resolution = resolve_gpu(gpu)
+    print_gpu_diagnostics(gpu_resolution)
+    if dry_run:
+        print(f"[EVAL-DRY-RUN] model={model} dataset={dataset} data_root={cfg['data_root']}")
+        return 0
+    ckpt = Path(checkpoint) if checkpoint else _best_checkpoint(model, dataset)
+    if not ckpt.is_absolute():
+        ckpt = ROOT / ckpt
+    print(f"[EVAL] model={model} dataset={dataset} checkpoint={ckpt}")
+    adapter = get_model_adapter(model)
+    if not adapter.supports_inprocess_eval:
+        raise ModelEvaluationUnavailable(adapter.notes_or_failure_reason)
+    device = torch.device(gpu_resolution.local_device)
+    _, code = evaluate_with_adapter(
+        model_name=model,
+        dataset_cfg=cfg,
+        model_config=load_model_config(model),
+        adapter=adapter,
+        checkpoint_path=ckpt,
+        device=device,
+        strict_profiling=not allow_missing_profilers,
+    )
+    return code
+
+
+def main() -> int:
+    args = _parse_args()
+    models = sorted(_load_registry()) if args.model == "all" else [args.model]
+    datasets = _load_datasets() if args.dataset == "all" else [args.dataset]
+    rows = []
+    for model in models:
+        for dataset in datasets:
+            try:
+                code = _evaluate_one(
+                    model,
+                    dataset,
+                    args.checkpoint,
+                    args.gpu,
+                    args.dry_run,
+                    args.allow_missing_profilers,
+                )
+                rows.append((model, dataset, "complete" if code == 0 else "failed"))
+            except FileNotFoundError as exc:
+                print(f"[MISSING] {model}/{dataset}: {exc}")
+                rows.append((model, dataset, "missing"))
+            except ModelEvaluationUnavailable as exc:
+                print(f"[UNAVAILABLE] {model}/{dataset}: {exc}")
+                rows.append((model, dataset, "unavailable"))
+            except RuntimeError as exc:
+                print(f"[FAILED] {model}/{dataset}: {exc}")
+                rows.append((model, dataset, "failed"))
+    append_to_comparison_table()
+    print("\nModel | Dataset | Status")
+    print("--- | --- | ---")
+    for model, dataset, status in rows:
+        print(f"{model} | {dataset} | {status}")
+    return 0 if all(status == "complete" for _, _, status in rows) else 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

generated_configs/dsifn_cd__bifa.json

@@ -0,0 +1,61 @@
+{
+  "name": "dsifn_cd-train-bifa",
+  "phase": "train",
+  "gpu_ids": [
+    0
+  ],
+  "path_cd": {
+    "log": "logs",
+    "result": "results",
+    "checkpoint": "checkpoint",
+    "resume_state": null
+  },
+  "datasets": {
+    "train": {
+      "name": "DSIFN-CD",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/dsifn_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": true,
+      "data_len": -1
+    },
+    "val": {
+      "name": "DSIFN-CD",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/dsifn_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    },
+    "test": {
+      "name": "DSIFN-CD",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/dsifn_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    }
+  },
+  "model": {
+    "name": "bifa",
+    "loss": "ce_dice"
+  },
+  "train": {
+    "n_epoch": 200,
+    "train_print_iter": 50,
+    "val_freq": 1,
+    "val_print_iter": 20,
+    "optimizer": {
+      "type": "adam",
+      "lr": 0.0001
+    },
+    "sheduler": {
+      "lr_policy": "linear",
+      "n_step": 3,
+      "gamma": 0.1
+    }
+  }
+}

generated_configs/dsifn_cd__cdmamba.json

@@ -0,0 +1,91 @@
+{
+  "name": "dsifn_cd-train-cdmamba",
+  "phase": "train",
+  "gpu_ids": [
+    0
+  ],
+  "path_cd": {
+    "log": "logs",
+    "result": "results",
+    "checkpoint": "checkpoint",
+    "resume_state": null
+  },
+  "datasets": {
+    "train": {
+      "name": "DSIFN-CD",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/dsifn_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 2,
+      "batch_size": 8,
+      "use_shuffle": true,
+      "data_len": -1
+    },
+    "val": {
+      "name": "DSIFN-CD",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/dsifn_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 2,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    },
+    "test": {
+      "name": "DSIFN-CD",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/dsifn_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 2,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    }
+  },
+  "model": {
+    "name": "cdmamba",
+    "loss": "ce_dice",
+    "init_filters": 16,
+    "n_classes": 2,
+    "mode": "AGLGF",
+    "conv_mode": "orignal_dinner",
+    "local_query_model": "orignal_dinner",
+    "up_mode": "SRCM",
+    "up_conv_mode": "deepwise",
+    "spatial_dims": 2,
+    "in_channels": 3,
+    "resdiual": false,
+    "blocks_down": [
+      1,
+      2,
+      2,
+      4
+    ],
+    "blocks_up": [
+      1,
+      1,
+      1
+    ],
+    "diff_abs": "later",
+    "stage": 2,
+    "mamba_act": "relu",
+    "norm": [
+      "GROUP",
+      {
+        "num_groups": 8
+      }
+    ]
+  },
+  "train": {
+    "n_epoch": 200,
+    "train_print_iter": 50,
+    "val_freq": 1,
+    "val_print_iter": 20,
+    "optimizer": {
+      "type": "adam",
+      "lr": 0.0001
+    },
+    "sheduler": {
+      "lr_policy": "linear",
+      "n_step": 3,
+      "gamma": 0.1
+    }
+  }
+}

generated_configs/dsifn_cd__changer_opencd.py

@@ -0,0 +1,26 @@
+_base_ = '/new-home/buddhiw/CD-Models/model_repos/open-cd/configs/changer/changer_ex_r18_512x512_40k_levircd.py'
+
+dataset_type = 'DSIFN_Dataset'
+data_root = r'/new-home/buddhiw/CD-Models/generated_dataset_views/dsifn_cd/opencd'
+crop_size = (256, 256)
+
+train_dataloader = dict(
+    batch_size=8,
+    num_workers=4,
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        data_prefix=dict(seg_map_path='train/label', img_path_from='train/A', img_path_to='train/B')))
+val_dataloader = dict(
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        data_prefix=dict(seg_map_path='val/label', img_path_from='val/A', img_path_to='val/B')))
+test_dataloader = dict(
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        data_prefix=dict(seg_map_path='test/label', img_path_from='test/A', img_path_to='test/B')))
+
+train_cfg = dict(type='IterBasedTrainLoop', max_iters=90000, val_interval=450)
+work_dir = r'/new-home/buddhiw/CD-Models/results/changer/dsifn_cd/work_dir'

generated_configs/dsifn_cd__hanet_metadata.json

@@ -0,0 +1,19 @@
+{
+  "patch_size": 256,
+  "augmentation": true,
+  "num_gpus": 1,
+  "num_workers": 4,
+  "num_channel": 3,
+  "EF": false,
+  "epochs": 200,
+  "epochs_threshold": 15,
+  "gamma": 0.5,
+  "weight_decay": 0.0001,
+  "batch_size": 8,
+  "learning_rate": 0.0001,
+  "loss_function": "hybrid",
+  "dataset_dir": "/new-home/buddhiw/CD-Models/generated_dataset_views/dsifn_cd/hanet_matched/",
+  "weight_dir": "/new-home/buddhiw/CD-Models/results/hanet/dsifn_cd/weights/",
+  "Output_dir": "/new-home/buddhiw/CD-Models/results/hanet/dsifn_cd/outputs/",
+  "log_dir": "/new-home/buddhiw/CD-Models/results/hanet/dsifn_cd/logs"
+}

generated_configs/kate_cd__bifa.json

@@ -0,0 +1,61 @@
+{
+  "name": "kate_cd-train-bifa",
+  "phase": "train",
+  "gpu_ids": [
+    0
+  ],
+  "path_cd": {
+    "log": "logs",
+    "result": "results",
+    "checkpoint": "checkpoint",
+    "resume_state": null
+  },
+  "datasets": {
+    "train": {
+      "name": "KATE-CD-256",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/kate_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": true,
+      "data_len": -1
+    },
+    "val": {
+      "name": "KATE-CD-256",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/kate_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    },
+    "test": {
+      "name": "KATE-CD-256",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/kate_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    }
+  },
+  "model": {
+    "name": "bifa",
+    "loss": "ce_dice"
+  },
+  "train": {
+    "n_epoch": 200,
+    "train_print_iter": 50,
+    "val_freq": 1,
+    "val_print_iter": 20,
+    "optimizer": {
+      "type": "adam",
+      "lr": 0.0001
+    },
+    "sheduler": {
+      "lr_policy": "linear",
+      "n_step": 3,
+      "gamma": 0.1
+    }
+  }
+}

generated_configs/levir_cd__bifa.json

@@ -0,0 +1,61 @@
+{
+  "name": "levir_cd-train-bifa",
+  "phase": "train",
+  "gpu_ids": [
+    0
+  ],
+  "path_cd": {
+    "log": "logs",
+    "result": "results",
+    "checkpoint": "checkpoint",
+    "resume_state": null
+  },
+  "datasets": {
+    "train": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": true,
+      "data_len": -1
+    },
+    "val": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    },
+    "test": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    }
+  },
+  "model": {
+    "name": "bifa",
+    "loss": "ce_dice"
+  },
+  "train": {
+    "n_epoch": 200,
+    "train_print_iter": 50,
+    "val_freq": 1,
+    "val_print_iter": 20,
+    "optimizer": {
+      "type": "adam",
+      "lr": 0.0001
+    },
+    "sheduler": {
+      "lr_policy": "linear",
+      "n_step": 3,
+      "gamma": 0.1
+    }
+  }
+}

generated_configs/levir_cd__cdmamba.json

@@ -0,0 +1,91 @@
+{
+  "name": "levir_cd-train-cdmamba",
+  "phase": "train",
+  "gpu_ids": [
+    0
+  ],
+  "path_cd": {
+    "log": "logs",
+    "result": "results",
+    "checkpoint": "checkpoint",
+    "resume_state": null
+  },
+  "datasets": {
+    "train": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 1,
+      "batch_size": 8,
+      "use_shuffle": true,
+      "data_len": -1
+    },
+    "val": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 1,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    },
+    "test": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 1,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    }
+  },
+  "model": {
+    "name": "cdmamba",
+    "loss": "ce_dice",
+    "init_filters": 16,
+    "n_classes": 2,
+    "mode": "AGLGF",
+    "conv_mode": "orignal_dinner",
+    "local_query_model": "orignal_dinner",
+    "up_mode": "SRCM",
+    "up_conv_mode": "deepwise",
+    "spatial_dims": 2,
+    "in_channels": 3,
+    "resdiual": false,
+    "blocks_down": [
+      1,
+      2,
+      2,
+      4
+    ],
+    "blocks_up": [
+      1,
+      1,
+      1
+    ],
+    "diff_abs": "later",
+    "stage": 2,
+    "mamba_act": "relu",
+    "norm": [
+      "GROUP",
+      {
+        "num_groups": 8
+      }
+    ]
+  },
+  "train": {
+    "n_epoch": 200,
+    "train_print_iter": 50,
+    "val_freq": 1,
+    "val_print_iter": 20,
+    "optimizer": {
+      "type": "adam",
+      "lr": 0.0001
+    },
+    "sheduler": {
+      "lr_policy": "linear",
+      "n_step": 3,
+      "gamma": 0.1
+    }
+  }
+}

generated_configs/levir_cd__changer_opencd.py

@@ -0,0 +1,26 @@
+_base_ = '/new-home/buddhiw/CD-Models/model_repos/open-cd/configs/changer/changer_ex_r18_512x512_40k_levircd.py'
+
+dataset_type = 'DSIFN_Dataset'
+data_root = r'/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd/opencd'
+crop_size = (256, 256)
+
+train_dataloader = dict(
+    batch_size=8,
+    num_workers=4,
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        data_prefix=dict(seg_map_path='train/label', img_path_from='train/A', img_path_to='train/B')))
+val_dataloader = dict(
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        data_prefix=dict(seg_map_path='val/label', img_path_from='val/A', img_path_to='val/B')))
+test_dataloader = dict(
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        data_prefix=dict(seg_map_path='test/label', img_path_from='test/A', img_path_to='test/B')))
+
+train_cfg = dict(type='IterBasedTrainLoop', max_iters=204000, val_interval=1020)
+work_dir = r'/new-home/buddhiw/CD-Models/results/changer/levir_cd/work_dir'

generated_configs/levir_cd__hanet_metadata.json

@@ -0,0 +1,19 @@
+{
+  "patch_size": 256,
+  "augmentation": true,
+  "num_gpus": 1,
+  "num_workers": 4,
+  "num_channel": 3,
+  "EF": false,
+  "epochs": 200,
+  "epochs_threshold": 15,
+  "gamma": 0.5,
+  "weight_decay": 0.0001,
+  "batch_size": 8,
+  "learning_rate": 0.0001,
+  "loss_function": "hybrid",
+  "dataset_dir": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd/hanet_matched/",
+  "weight_dir": "/new-home/buddhiw/CD-Models/results/hanet/levir_cd/weights/",
+  "Output_dir": "/new-home/buddhiw/CD-Models/results/hanet/levir_cd/outputs/",
+  "log_dir": "/new-home/buddhiw/CD-Models/results/hanet/levir_cd/logs"
+}

generated_configs/levir_cd_test_as_val__bifa.json

@@ -0,0 +1,61 @@
+{
+  "name": "levir_cd_test_as_val-train-bifa",
+  "phase": "train",
+  "gpu_ids": [
+    0
+  ],
+  "path_cd": {
+    "log": "logs",
+    "result": "results",
+    "checkpoint": "checkpoint",
+    "resume_state": null
+  },
+  "datasets": {
+    "train": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd_test_as_val/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": true,
+      "data_len": -1
+    },
+    "val": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd_test_as_val/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    },
+    "test": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd_test_as_val/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    }
+  },
+  "model": {
+    "name": "bifa",
+    "loss": "ce_dice"
+  },
+  "train": {
+    "n_epoch": 200,
+    "train_print_iter": 50,
+    "val_freq": 1,
+    "val_print_iter": 20,
+    "optimizer": {
+      "type": "adam",
+      "lr": 0.0001
+    },
+    "sheduler": {
+      "lr_policy": "linear",
+      "n_step": 3,
+      "gamma": 0.1
+    }
+  }
+}

generated_configs/levir_cd_test_as_val__cdmamba.json

@@ -0,0 +1,91 @@
+{
+  "name": "levir_cd_test_as_val-train-cdmamba",
+  "phase": "train",
+  "gpu_ids": [
+    0
+  ],
+  "path_cd": {
+    "log": "logs",
+    "result": "results",
+    "checkpoint": "checkpoint",
+    "resume_state": null
+  },
+  "datasets": {
+    "train": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd_test_as_val/legacy_list",
+      "resolution": 256,
+      "num_workers": 1,
+      "batch_size": 8,
+      "use_shuffle": true,
+      "data_len": -1
+    },
+    "val": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd_test_as_val/legacy_list",
+      "resolution": 256,
+      "num_workers": 1,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    },
+    "test": {
+      "name": "LEVIR-CD+",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd_test_as_val/legacy_list",
+      "resolution": 256,
+      "num_workers": 1,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    }
+  },
+  "model": {
+    "name": "cdmamba",
+    "loss": "ce_dice",
+    "init_filters": 16,
+    "n_classes": 2,
+    "mode": "AGLGF",
+    "conv_mode": "orignal_dinner",
+    "local_query_model": "orignal_dinner",
+    "up_mode": "SRCM",
+    "up_conv_mode": "deepwise",
+    "spatial_dims": 2,
+    "in_channels": 3,
+    "resdiual": false,
+    "blocks_down": [
+      1,
+      2,
+      2,
+      4
+    ],
+    "blocks_up": [
+      1,
+      1,
+      1
+    ],
+    "diff_abs": "later",
+    "stage": 2,
+    "mamba_act": "relu",
+    "norm": [
+      "GROUP",
+      {
+        "num_groups": 8
+      }
+    ]
+  },
+  "train": {
+    "n_epoch": 200,
+    "train_print_iter": 50,
+    "val_freq": 1,
+    "val_print_iter": 20,
+    "optimizer": {
+      "type": "adam",
+      "lr": 0.0001
+    },
+    "sheduler": {
+      "lr_policy": "linear",
+      "n_step": 3,
+      "gamma": 0.1
+    }
+  }
+}

generated_configs/levir_cd_test_as_val__changer_opencd.py

@@ -0,0 +1,26 @@
+_base_ = '/new-home/buddhiw/CD-Models/model_repos/open-cd/configs/changer/changer_ex_r18_512x512_40k_levircd.py'
+
+dataset_type = 'DSIFN_Dataset'
+data_root = r'/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd_test_as_val/opencd'
+crop_size = (256, 256)
+
+train_dataloader = dict(
+    batch_size=8,
+    num_workers=4,
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        data_prefix=dict(seg_map_path='train/label', img_path_from='train/A', img_path_to='train/B')))
+val_dataloader = dict(
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        data_prefix=dict(seg_map_path='val/label', img_path_from='val/A', img_path_to='val/B')))
+test_dataloader = dict(
+    dataset=dict(
+        type=dataset_type,
+        data_root=data_root,
+        data_prefix=dict(seg_map_path='test/label', img_path_from='test/A', img_path_to='test/B')))
+
+train_cfg = dict(type='IterBasedTrainLoop', max_iters=254800, val_interval=1274)
+work_dir = r'/new-home/buddhiw/CD-Models/results/changer/levir_cd_test_as_val/work_dir'

generated_configs/levir_cd_test_as_val__hanet_metadata.json

@@ -0,0 +1,19 @@
+{
+  "patch_size": 256,
+  "augmentation": true,
+  "num_gpus": 1,
+  "num_workers": 4,
+  "num_channel": 3,
+  "EF": false,
+  "epochs": 200,
+  "epochs_threshold": 15,
+  "gamma": 0.5,
+  "weight_decay": 0.0001,
+  "batch_size": 8,
+  "learning_rate": 0.0001,
+  "loss_function": "hybrid",
+  "dataset_dir": "/new-home/buddhiw/CD-Models/generated_dataset_views/levir_cd_test_as_val/hanet_matched/",
+  "weight_dir": "/new-home/buddhiw/CD-Models/results/hanet/levir_cd_test_as_val/weights/",
+  "Output_dir": "/new-home/buddhiw/CD-Models/results/hanet/levir_cd_test_as_val/outputs/",
+  "log_dir": "/new-home/buddhiw/CD-Models/results/hanet/levir_cd_test_as_val/logs"
+}

generated_configs/sysu_cd__bifa.json

@@ -0,0 +1,61 @@
+{
+  "name": "sysu_cd-train-bifa",
+  "phase": "train",
+  "gpu_ids": [
+    0
+  ],
+  "path_cd": {
+    "log": "logs",
+    "result": "results",
+    "checkpoint": "checkpoint",
+    "resume_state": null
+  },
+  "datasets": {
+    "train": {
+      "name": "SYSU-CD",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/sysu_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": true,
+      "data_len": -1
+    },
+    "val": {
+      "name": "SYSU-CD",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/sysu_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    },
+    "test": {
+      "name": "SYSU-CD",
+      "datasetroot": "/new-home/buddhiw/CD-Models/generated_dataset_views/sysu_cd/legacy_list",
+      "resolution": 256,
+      "num_workers": 4,
+      "batch_size": 8,
+      "use_shuffle": false,
+      "data_len": -1
+    }
+  },
+  "model": {
+    "name": "bifa",
+    "loss": "ce_dice"
+  },
+  "train": {
+    "n_epoch": 200,
+    "train_print_iter": 50,
+    "val_freq": 1,
+    "val_print_iter": 20,
+    "optimizer": {
+      "type": "adam",
+      "lr": 0.0001
+    },
+    "sheduler": {
+      "lr_policy": "linear",
+      "n_step": 3,
+      "gamma": 0.1
+    }
+  }
+}