Research map / Video / Action recognition
SKiT: a Fast Key Information Video Transformer for Online Surgical Phase Recognition
Yang Liu, Jiayu Huo, Jingjing Peng, Rachel Sparks, Prokar Dasgupta, Alejandro Granados, Sebastien Ourselin
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 Key-recorder global key feature to local features (ablation) | SKiT · Cholec80 | accuracy: 89.3 -> 92.5 | Improved |
| record key feature from spatial features instead of local temporal features (ablation) | SKiT · Cholec80 | accuracy: 92.5 -> 88.1 | Hurt |
| use Key-recorder instead of LSTM (ablation) | SKiT · Cholec80 | accuracy: 89.5 -> 92.5 | Improved |
| use Key-recorder instead of TCN (ablation) | SKiT · Cholec80 | accuracy: 90.1 -> 92.5 | Improved |
| use Key-recorder instead of GRU (ablation) | SKiT · Cholec80 | accuracy: 89.6 -> 92.5 | Improved |
| increase key feature length from 8 to 64 (ablation) | SKiT · Cholec80 | accuracy: 90.6 -> 92.5 | Improved |
| increase key feature length from 64 to 128 (ablation) | SKiT · Cholec80 | accuracy: 92.5 -> 92.1 | Hurt |
| increase key feature length from 8 to 32 (ablation) | SKiT · AutoLaparo | accuracy: 82.6 -> 82.9 | Mixed |
| vary local window size lambda from 20 to 200 (ablation) | SKiT · Cholec80 | accuracy: 92.5 -> 92.3 | No clear effect |
| increase input length T from 100 to 3000 (ablation) | SKiT · Cholec80 | accuracy: 90.2 -> 92.5 | Improved |
| replace LoViT global ProbSparse aggregator with key pooling | SKiT vs LoViT · Cholec80 | accuracy: 91.5 -> 92.5 | Improved |
| replace LoViT global ProbSparse aggregator with key pooling | SKiT vs LoViT · AutoLaparo | accuracy: 81.4 -> 82.9 | Improved |
Closest papers by meaning
- 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)
- CholecTrack20: A Multi-Perspective Tracking Dataset for Surgical Tools (CVPR 2025 · similarity 0.83)
- SurgCoT: Advancing Spatiotemporal Reasoning in Surgical Videos through a Chain-of-Thought Benchmark (CVPR 2026 · similarity 0.83)
- Bridging Vision and Language for Robust Context-Aware Surgical Point Tracking: The VL-SurgPT Dataset and Benchmark (AAAI 2026 · similarity 0.83)
- SurgicalSAM: Efficient Class Promptable Surgical Instrument Segmentation (AAAI 2024 · similarity 0.82)
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.
