Clean model card: one official checkpoint and result per dataset

README.md

@@ -9,21 +9,27 @@ tags:
   - competitorformer
 ---
 
-# CompetitorFormer Checkpoints
+# CompetitorFormer
 
-Official weights for **CompetitorFormer** (CVPR 2026).
+> **CompetitorFormer: Mitigating Query Conflicts for 3D Instance Segmentation via Competitive Strategy (CVPR 2026)**
+>
+> Duanchu Wang, Junjie Yang, Haoran Gong, Jing Liu, Di Wang
+>
+> [[Paper](https://openaccess.thecvf.com/content/CVPR2026/html/Wang_CompetitorFormer_Mitigating_Query_Conflicts_for_3D_Instance_Segmentation_via_Competitive_CVPR_2026_paper.html)]
+> [[Code](https://github.com/DuanchuWang/CompetitorFormer)]
 
-Code: [https://github.com/DuanchuWang/CompetitorFormer](https://github.com/DuanchuWang/CompetitorFormer)
+Official checkpoints for CompetitorFormer, a competitive-strategy-based transformer that mitigates query conflicts for 3D instance segmentation.
 
-## ScanNet++ fair val
+## Results & Models
 
-| File | Original training save | AP / AP50 / AP25 | `num_query` |
-| --- | --- | --- | ---: |
-| [`competitorformer_scannetpp.pth`](https://huggingface.co/WangDuanchu/CompetitorFormer/resolve/main/competitorformer_scannetpp.pth) | `epoch490_AP_0.3335_0.4725_0.5640.pth` | **0.3335 / 0.4725 / 0.5640** | 500 |
+### ScanNet++ (official 50-scene val)
 
-This is the official ScanNet++ fair val checkpoint (50-scene `nvs_sem_val` split).
+- Model: CompetitorFormer on ScanNet++ official 50-scene val
+- Checkpoint: [`competitorformer_scannetpp.pth`](https://huggingface.co/WangDuanchu/CompetitorFormer/resolve/main/competitorformer_scannetpp.pth) (`num_query=500`)
+- AP / AP<sub>50</sub> / AP<sub>25</sub> = **0.339 / 0.485 / 0.581**
 
 ```bash
+mkdir -p checkpoints
 wget https://huggingface.co/WangDuanchu/CompetitorFormer/resolve/main/competitorformer_scannetpp.pth \
   -O checkpoints/competitorformer_scannetpp.pth
 ```
@@ -34,10 +40,44 @@ or:
 huggingface-cli download WangDuanchu/CompetitorFormer competitorformer_scannetpp.pth --local-dir checkpoints
 ```
 
-## ScanNet v2 fair val (optional)
+```bash
+python tools/train.py configs/scannetpp/competitorformer_scannetpp.yaml \
+  --work_dir exps/competitorformer_scannetpp \
+  --eval_only \
+  --resume checkpoints/competitorformer_scannetpp.pth
+```
+
+Inference knobs (do not reuse ScanNet v2 values): `num_query=500`, `topk_insts=1300`, `num_class=84`.
+
+### ScanNet v2
+
+- Checkpoint: [`competitorformer_scannet.pth`](https://huggingface.co/WangDuanchu/CompetitorFormer/resolve/main/competitorformer_scannet.pth) (`num_query=400`)
+
+```bash
+wget https://huggingface.co/WangDuanchu/CompetitorFormer/resolve/main/competitorformer_scannet.pth \
+  -O checkpoints/competitorformer_scannet.pth
+```
+
+or:
+
+```bash
+huggingface-cli download WangDuanchu/CompetitorFormer competitorformer_scannet.pth --local-dir checkpoints
+```
+
+See the [GitHub README](https://github.com/DuanchuWang/CompetitorFormer) for installation, data preparation, training, and evaluation.
+
+## Citation
+
+```bibtex
+@InProceedings{Wang_2026_CVPR,
+  title     = {CompetitorFormer: Mitigating Query Conflicts for 3D Instance Segmentation via Competitive Strategy},
+  author    = {Wang, Duanchu and Yang, Junjie and Gong, Haoran and Liu, Jing and Wang, Di},
+  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
+  pages     = {34724-34733},
+  year      = {2026},
+}
+```
 
-| File | Original training save | AP / AP50 / AP25 | `num_query` |
-| --- | --- | --- | ---: |
-| [`competitorformer_scannet.pth`](https://huggingface.co/WangDuanchu/CompetitorFormer/resolve/main/competitorformer_scannet.pth) | `epoch440_AP_0.6276_0.8072_0.8718.pth` | **0.6276 / 0.8072 / 0.8718** | 400 |
+## License
 
-See the GitHub README for evaluation commands and data preparation.
+This project is released under the MIT License.