Promote convnext_tiny_all_lands_r8_s67 (nonland macro F1 0.611)

README.md

@@ -1,3 +1,94 @@
----
-license: mit
----
+---
+tags:
+- image-classification
+- multi-label-classification
+- timm
+- pytorch
+---
+
+<!-- Generated by scripts/promote_model.py on every promotion: edits made here are overwritten. -->
+
+# MTG Color Classifier
+
+Looks at a piece of Magic: The Gathering card art and predicts which of the five
+Magic colors it belongs to: **W**hite, bl**U**e, **B**lack, **R**ed, **G**reen. It is
+multi-label: each color gets its own probability and its own decision threshold,
+so a picture can be several colors, or none (colorless). The model only ever sees
+the image, never the card's name, text or type.
+
+Trained by fine-tuning an ImageNet-pretrained `convnext_tiny.fb_in22k_ft_in1k` (via `timm`) on
+the art crops of every unique illustration on [Scryfall](https://scryfall.com).
+Lands are labeled with their color identity (Forest -> G, a dual land -> both colors); every other card with its `colors` field.
+
+## Current model
+
+| | |
+|---|---|
+| Run | `convnext_tiny_all_lands_r8_s67` |
+| Backbone | `convnext_tiny.fb_in22k_ft_in1k` |
+| Test macro F1 (nonland test cards, n=4,477) | **0.611** |
+| Test macro AP | 0.657 |
+| Exact match (all five colors right) | 0.416 |
+| Promoted | 2026-10-19T20:01:47+00:00 |
+| Trained with | torch 2.11.0+cu128, timm 1.0.29 |
+
+| Color | Test F1 | Threshold |
+|---|---|---|
+| W | 0.567 | 0.25 |
+| U | 0.621 | 0.42 |
+| B | 0.548 | 0.37 |
+| R | 0.643 | 0.31 |
+| G | 0.674 | 0.34 |
+
+Thresholds were tuned per color on the validation split. Test cards never share
+an `oracle_id` with training cards, so the scores are on cards the model never saw.
+
+## Files
+
+- `model_bundle.pt`: one dict with the weights (`state_dict`) and everything needed
+  to use them: `backbone`, `classes`, `thresholds`, `preprocess`. Loads with
+  `torch.load(..., weights_only=True)`.
+- `model_info.json`: the same fields without the weights.
+
+## Usage
+
+```python
+import timm, torch
+from huggingface_hub import hf_hub_download
+from io import BytesIO
+from PIL import Image
+from torchvision import transforms as T
+
+bundle = torch.load(hf_hub_download("paularoni/mtg-color-classifier", "model_bundle.pt"),
+                    map_location="cpu", weights_only=True)
+net = timm.create_model(bundle["backbone"], pretrained=False, num_classes=len(bundle["classes"]))
+net.load_state_dict(bundle["state_dict"])
+net.eval()
+
+pp = bundle["preprocess"]
+im = Image.open("art.jpg").convert("RGB")  # ideally color-managed to sRGB first
+im = im.resize((pp["cache_size"], pp["cache_size"]), Image.Resampling.LANCZOS)
+buf = BytesIO()  # JPEG round trip: the training images were JPEGs
+im.save(buf, "JPEG", quality=pp["cache_jpeg_quality"], subsampling=pp["cache_jpeg_subsampling"])
+im = Image.open(buf).convert("RGB")
+x = T.Compose([T.Resize((pp["img_size"], pp["img_size"])), T.ToTensor(),
+               T.Normalize(pp["mean"], pp["std"])])(im)
+with torch.no_grad():
+    probs = torch.sigmoid(net(x.unsqueeze(0)))[0].tolist()
+colors = [c for c, p, t in zip(bundle["classes"], probs, bundle["thresholds"]) if p >= t]
+print(dict(zip(bundle["classes"], probs)), "".join(colors) or "colorless")
+```
+
+Images are squashed to a square (no crop): color cues are often at the edges of the art.
+The two-step resize (Lanczos to 256, JPEG quality 95, then bilinear
+to 224) matches how the training images were prepared. Skipping the JPEG step
+shifts some probabilities by up to 0.1.
+
+## Limitations
+
+- Art alone is often ambiguous: a macro F1 around 0.61 means many pictures get
+  a wrong or missing color. Treat the probabilities as a guess, not a rule lookup.
+- Trained on Magic card art; other pictures get an answer, but it means little.
+
+Unofficial fan project, not affiliated with or endorsed by Wizards of the Coast.
+Card art is (c) Wizards of the Coast and its artists; card data and images from Scryfall.

model_bundle.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cb06ee58206ac1f398f03bf2efd969ed86798578361062ee7f13ba773b45119
+size 111354175

model_info.json

@@ -0,0 +1,65 @@
+{
+  "format_version": 1,
+  "run": "convnext_tiny_all_lands_r8_s67",
+  "backbone": "convnext_tiny.fb_in22k_ft_in1k",
+  "classes": [
+    "W",
+    "U",
+    "B",
+    "R",
+    "G"
+  ],
+  "thresholds": [
+    0.25,
+    0.42,
+    0.37,
+    0.31,
+    0.34
+  ],
+  "preprocess": {
+    "resize": "squash",
+    "color": "srgb",
+    "cache_size": 256,
+    "cache_resample": "lanczos",
+    "cache_jpeg_quality": 95,
+    "cache_jpeg_subsampling": "4:4:4",
+    "img_size": 224,
+    "resample": "bilinear",
+    "mean": [
+      0.485,
+      0.456,
+      0.406
+    ],
+    "std": [
+      0.229,
+      0.224,
+      0.225
+    ]
+  },
+  "land_labels": "identity",
+  "metrics": {
+    "population": "nonland test cards",
+    "n": 4477,
+    "macro_f1": 0.6105,
+    "macro_ap": 0.6575,
+    "f1": {
+      "W": 0.5667,
+      "U": 0.6207,
+      "B": 0.548,
+      "R": 0.6434,
+      "G": 0.6738
+    },
+    "exact_match": 0.4157
+  },
+  "training": {
+    "torch": "2.11.0+cu128",
+    "timm": "1.0.29",
+    "run_tag": "r8",
+    "land_variant": "all_lands",
+    "seed": 67,
+    "best_val_macro_ap": 0.6556,
+    "kaggle_notebook": "paularoni/mtg-color-classification-training"
+  },
+  "weights_sha256": "a00d794b7397f7137cb6299d02c7707a19a5b31c90e6b74ad274f4591c4e4d68",
+  "promoted_at": "2026-10-19T20:01:47+00:00"
+}