Research map / Video / Video-language understanding
Mistake Attribution: Fine-Grained Mistake Understanding in Egocentric Videos
Yayuan Li, Aadit Jain, Filippos Bellos, Jason J. Corso
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 |
|---|---|---|---|
| use MisFormer instead of best video-language baseline | semantic attribution · EPIC-KITCHENS-M | avg F1: 77.23 -> 83.89 | Improved |
| use MisFormer instead of best baseline | semantic attribution · Ego4D-M | avg F1: 50.95 -> 56.24 | Improved |
| use MisFormer instead of EgoT2 | temporal attribution (PNR) · Ego4D-M | MAE frames: 24.48 -> 19.14 | Improved |
| use MisFormer instead of SSDA hand-object detector | spatial attribution · Ego4D-M | mIoU: 64.54 -> 59.21 | Hurt |
| use MisFormer instead of hand-detection MediaPipe-U | spatial attribution · Ego4D-M | mIoU: 49.88 -> 59.21 | Improved |
| use MisFormer instead of EgoPED | mistake detection · EPIC-KITCHENS-M | F1@0.5: 21.18 -> 78.05 | Improved |
| train MisFormer from scratch on EgoPER | mistake detection · EgoPER | F1@0.5: 32.18 (EgoPED) vs 26.78 | Hurt |
| pretrain MisFormer on EPIC-KITCHENS-M before EgoPER fine-tuning (ablation) | MisFormer · EgoPER | F1@0.5: 26.78 -> 35.18 | Improved |
| use InternVideo2 instead of LaViLa features (ablation) | MisFormer · Ego4D-M | semantic F1: 49.16 -> 56.24 | Improved |
| add projection block P (ablation) | MisFormer · Ego4D-M | semantic F1: 51.34 -> 56.24 | Improved |
| add temporal attribution supervision (ablation) | MisFormer · Ego4D-M | temporal MAE (s): 0.623 -> 0.438 | Improved |
| use attention-based heatmap instead of GradCAM (ablation) | MisFormer · Ego4D-M | spatial mIoU: 55.03 -> 59.21 | Improved |
Closest papers by meaning
- Gazing Into Missteps: Leveraging Eye-Gaze for Unsupervised Mistake Detection in Egocentric Videos of Skilled Human Activities (CVPR 2025 · similarity 0.89)
- PREGO: Online Mistake Detection in PRocedural EGOcentric Videos (CVPR 2024 · similarity 0.86)
- Error Detection in Egocentric Procedural Task Videos (CVPR 2024 · similarity 0.85)
- Test-Time Adaptation for Combating Missing Modalities in Egocentric Videos (ICLR 2025 · similarity 0.85)
- MMEgo: Towards Building Egocentric Multimodal LLMs for Video QA (ICLR 2025 · similarity 0.84)
- EgoThinker: Unveiling Egocentric Reasoning with Spatio-Temporal CoT (NeurIPS 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.
