Upload folder using huggingface_hub

README.md

@@ -0,0 +1,65 @@
+---
+license: cc-by-4.0
+library_name: transformers
+tags:
+  - surgical-video
+  - spatio-temporal-grounding
+  - medical-vision-language-model
+  - eccv-2026
+---
+
+# RefineRank: Joint Box Refinement and Ranking for Surgical Spatio-Temporal Grounding
+
+Official RefineNet weights for the ECCV 2026 MedVidU Workshop paper
+**RefineRank**.
+
+RefineRank couples two **frozen** backbones — a MedVLM (Qwen2.5-VL
+architecture) and GroundingDINO — with a compact **1.25M-parameter** trainable
+module, **RefineNet** (`QueryConditionedProposalAdapter`). RefineNet uses
+MedVLM language and regional features to predict coordinate corrections and
+box-quality scores for GroundingDINO proposals; a parameter-free decoder then
+returns the highest-scoring original or refined box.
+
+- **Code, training & inference guide**: https://github.com/linzhe001/RefineRank
+- **Headline result**: 0.421 STG mIoU on the archived MedVidBench Community
+  leaderboard snapshot (27 July 2026) — the best STG mIoU among the ten
+  ranking metrics on that snapshot.
+- **Controlled evaluation** (video-separated split over CholecTrack20 /
+  CoPESD / EgoSurgery): STG mIoU 0.2719 → 0.4534 over the frozen
+  MedVLM + GroundingDINO baseline.
+
+## Files
+
+| File | Description |
+|---|---|
+| `proposal_adapter_full.pt` | Trained RefineNet checkpoint (`model_state`, `adapter_config`, `train_config`, `standardizer`, `source_feature_contract_hash`). SHA-256: `932e479b854c3d5fbafee25a3fcf9e6481e864fe98a6867502d6ebff39463d9b` |
+| `deployment_manifest.json` | Deployment contract used by `interface.py predict` (model kind, feature-contract hash, decoder mode) |
+
+This checkpoint is the exact `run_iter132_submission` run behind the paper's
+MedVidBench submission. The frozen MedVLM and GroundingDINO weights are
+third-party artifacts — see the GitHub README for where to obtain them.
+
+## Usage
+
+```bash
+git clone https://github.com/linzhe001/RefineRank
+cd RefineRank
+pip install -r requirements.txt
+# place proposal_adapter_full.pt + deployment_manifest.json under
+# checkpoints/refinenet/run_iter132_submission/ (see checkpoints/README.md)
+python interface.py predict --checkpoint checkpoints/refinenet/run_iter132_submission
+```
+
+## Citation
+
+```bibtex
+@inproceedings{jiang2026refinerank,
+  title     = {RefineRank: Joint Box Refinement and Ranking for Surgical
+               Spatio-Temporal Grounding},
+  author    = {Jiang, Linzhe and Huang, Jiayuan and Zhang, Changhao and
+               Jiang, Chunyang and Mao, Zhehua and
+               Garcia-Peraza-Herrera, Luis C. and Hoque, Mobarak I.},
+  booktitle = {ECCV Workshops (MedVidU)},
+  year      = {2026}
+}
+```

deployment_manifest.json

@@ -0,0 +1,17 @@
+{
+  "schema_version": 1,
+  "decision": "PROPOSAL_REPAIR_FULL_FIT_COMPLETE",
+  "model_kind": "proposal_repair_quality",
+  "selector_kind": "repair_quality",
+  "repair_variant": "full",
+  "query_mode": "q_last",
+  "decoder_mode": "fixed_current",
+  "operator_authorized": true,
+  "fit_protocol": "fixed_fold_selected_then_all_train_groups_full_fit",
+  "feature_contract_hash": "7fd672157bfa0174e802e66d1c37b5dd7996214edb191f9fba0c926f73d38748",
+  "proposal_checkpoint": "checkpoints/proposal_adapter_full.pt",
+  "proposal_checkpoint_sha256": "932e479b854c3d5fbafee25a3fcf9e6481e864fe98a6867502d6ebff39463d9b",
+  "frozen_inputs_unchanged": true,
+  "official_leaderboard_claim": false,
+  "release_note": "Camera-ready release manifest reconstructed from the checkpoint metadata (adapter_config, train_config, standardizer, source_feature_contract_hash). Original training run: runs/kept_wf10_final/iter132_submission in the SurgST-VLM research workspace."
+}

proposal_adapter_full.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:932e479b854c3d5fbafee25a3fcf9e6481e864fe98a6867502d6ebff39463d9b
+size 5013971