docs: sync model card from GitHub 23f150e

README.md

@@ -3,10 +3,10 @@ license: cc-by-nc-4.0
 library_name: pytorch
 pipeline_tag: image-feature-extraction
 tags:
-- person-re-identification
-- aerial-ground
-- vision-transformer
-- neurips2026
+  - person-re-identification
+  - aerial-ground
+  - vision-transformer
+  - neurips2026
 ---
 
 # 3D-VITAL
@@ -16,18 +16,18 @@ Cross-View Object Re-Identification** (NeurIPS 2026): [arXiv:2610.16692](https:/
 [poster](https://neurips.cc/virtual/2026/loc/paris/poster/150244).
 Code, training and evaluation are at <https://github.com/TurtleSmoke/3D-VITAL>.
 
-Each checkpoint is a ViT-B/16 + BNNeck trained on one camera type of AG-ReID, together with its 3D-VITAL synthetic
-views. Retrieval uses the cosine distance on the BNNeck feature, with a single forward pass.
-
 ## Checkpoints
 
-| File | Setting | Protocol 1 mAP / R1 | Protocol 2 mAP / R1 |
-|------|---------|---------------------|---------------------|
-| `agreid_aerial.pth` | AG-ReID, aerial training | A→G 52.09 / 66.20 | G→A 53.30 / 62.68 |
-| `agreid_ground.pth` | AG-ReID, ground training | A→G 55.99 / 67.13 | G→A 57.93 / 68.61 |
+Each checkpoint is trained on one camera type of AG-ReID, given by its name.
+
+| Checkpoint          | Query → Gallery | mAP   | Rank-1 |
+|---------------------|-----------------|-------|--------|
+| `agreid_aerial.pth` | A → G           | 52.09 | 66.20  |
+|                     | G → A           | 53.30 | 62.68  |
+| `agreid_ground.pth` | A → G           | 55.99 | 67.13  |
+|                     | G → A           | 57.93 | 68.61  |
 
-The numbers are those obtained with these files and `scripts/eval.sh`. The checkpoints keep only the inference
-parameters (backbone and BNNeck) and are stored in fp16, so they can differ from the paper by up to 0.1 point.
+A = aerial, G = ground. The checkpoints are stored in fp16, so they can differ slightly from the paper.
 
 ## Usage
 
@@ -37,8 +37,6 @@ hf download TurtleSmoke/3D-VITAL --include "*.pth" --local-dir checkpoints
 bash scripts/eval.sh agreid_aerial $DATA checkpoints/agreid_aerial.pth
 ```
 
-`$DATA` holds the real datasets, see the [GitHub README](https://github.com/TurtleSmoke/3D-VITAL#data).
-
 ## License
 
 The weights are released for non-commercial research under CC BY-NC 4.0. They were trained on
@@ -50,9 +48,9 @@ respective terms also apply. The code is under the MIT license.
 
 ```bibtex
 @inproceedings{grolleau2026vital,
-  title     = {3D-VITAL: Visibility-aware Identity Training from Artificial Liftings of 3D Reconstruction model for Cross-View Object Re-Identification},
-  author    = {Grolleau, William and Sabourin, Astrid and Lapouge, Guillaume and Achard, Catherine},
-  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
-  year      = {2026}
+    title = {3D-VITAL: Visibility-aware Identity Training from Artificial Liftings of 3D Reconstruction model for Cross-View Object Re-Identification},
+    author = {Grolleau, William and Sabourin, Astrid and Lapouge, Guillaume and Achard, Catherine},
+    booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
+    year = {2026}
 }
 ```