Research map / Video / Action recognition
A Stitch in Time: Learning Procedural Workflow via Self-Supervised Plackett-Luce Ranking
Chengan Che, Chao Wang, Xinyue Chen, Sophia Tsoka, Luis C. Garcia-Peraza-Herrera
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Plackett-Luce temporal ranking loss L_vid to MIM (ablation) | iBOT-style MIM baseline (ViT-S/16) · Cholec80 | k-NN accuracy: 69.4 -> 78.9 | Improved |
| add spatio-temporal jigsaw loss to MIM (ablation) | MIM baseline (ViT-S/16) · Cholec80 | k-NN accuracy: 69.4 -> 71.4 | Improved |
| add jigsaw on top of MIM + L_vid (ablation) | PL-Stitch (ViT-S/16) · Cholec80 | k-NN accuracy: 78.9 -> 80.2 | Improved |
| add jigsaw on top of MIM + L_vid (ablation) | PL-Stitch (ViT-S/16) · Cholec80 | linear probing accuracy: 77.1 -> 77.8 | Improved |
| use PL listwise ranking instead of pairwise loss (ablation) | L_vid (ViT-S/16) · Cholec80 | k-NN accuracy: 75.4 -> 78.9 | Improved |
| use PL listwise ranking instead of permutation classification (ablation) | L_vid (ViT-S/16) · Cholec80 | k-NN accuracy: 70.1 -> 78.9 | Improved |
| increase ranked frames k from 4 to 8 (ablation) | L_vid (ViT-S/16) · Cholec80 | k-NN accuracy: 78.1 -> 80.2 | Improved |
| increase ranked frames k from 8 to 16 (ablation) | L_vid (ViT-S/16) · Cholec80 | k-NN accuracy: 80.2 -> 80.3 | No clear effect |
| replace iBOT objective with PL-Stitch (PL ranking + jigsaw) | ViT-B/16 SSL pretraining on LEMON · Cholec80 | k-NN accuracy: 70.3 -> 81.7 | Improved |
| replace DINO with PL-Stitch pretraining | ViT-B/16 on Breakfast · Breakfast | linear probing accuracy: 15.9 -> 21.6 | Improved |
| replace best baseline with PL-Stitch | ViT-B/16 on GTEA · GTEA | k-NN accuracy: 60.0 -> 62.4 | Improved |
Closest papers by meaning
- CholecTrack20: A Multi-Perspective Tracking Dataset for Surgical Tools (CVPR 2025 · similarity 0.86)
- CaptainCook4D: A Dataset for Understanding Errors in Procedural Activities (NeurIPS 2024 · similarity 0.85)
- SurgCoT: Advancing Spatiotemporal Reasoning in Surgical Videos through a Chain-of-Thought Benchmark (CVPR 2026 · similarity 0.85)
- Recognize Any Surgical Object: Unleashing the Power of Weakly-Supervised Data (ICLR 2025 · similarity 0.84)
- Procedure-Aware Surgical Video-language Pretraining with Hierarchical Knowledge Augmentation (NeurIPS 2024 · similarity 0.84)
- VEDIT: Latent Prediction Architecture For Procedural Video Representation Learning (ICLR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
