Research map / Video / Action recognition
MistSense: Versatile Online Detection of Procedural and Execution Mistakes
Constantin Patsch, Yuankai Wu, Marsil Zakour, Driton Salihu, Eckehard Steinbach
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 hand pose modality via separate Hand Q-Former | MistSense RGB · Epic-tent-O | F1-score: 59.8 -> 63.3 | Improved |
| add predicted hand pose to RGB | MistSense · Holo Assist | F1-score: 54.8 -> 55.7 | Improved |
| use separate Q-Formers instead of a shared Video Q-Former for RGB and hand (ablation) | MistSense RGB+H(P) · Epic-tent-O | F1-score: 60.8 -> 63.3 | Improved |
| add Hand Q-Former instead of late fusion (ablation) | MistSense RGB+H(P) · Epic-tent-O | F1-score: 57.8 -> 63.3 | Improved |
| add RGB features to hand-only input (ablation) | MistSense · Epic-tent-O | F1-score: 49.1 -> 63.3 | Improved |
| late fusion of hand features versus RGB only (ablation) | MistSense · Epic-tent-O | F1-score: 59.8 -> 57.8 | No clear effect |
| add hand pose to RGB | MistSense · Holo Assist | mistake recall: 21.1 -> 31.6 | Improved |
| add hand pose to RGB | MistSense · Epic-tent-O | mistake precision: 55.0 -> 45.4 | Mixed |
| use MistSense RGB instead of DO | DO baseline · Epic-tent-O | F1-score: 46.5 -> 59.8 | Improved |
| use MistSense RGB instead of DO | DO baseline · Epic-tent-O | mistake F1: 23.7 -> 29.8 | Improved |
| use MistSense RGB instead of GazeCompl, no eye gaze | GazeCompl · Holo Assist | F1-score: 51.0 -> 54.8 | Improved |
| use MistSense explanation module instead of Video-ChatGPT | Video-ChatGPT · Epic-tent | CIDEr: 2.61 -> 3.09 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- PREGO — PREGO: Online Mistake Detection in PRocedural EGOcentric Videos (CVPR 2024) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PREGO: Online Mistake Detection in PRocedural EGOcentric Videos (CVPR 2024 · similarity 0.87)
- Error Detection in Egocentric Procedural Task Videos (CVPR 2024 · similarity 0.86)
- Modeling Multiple Normal Action Representations for Error Detection in Procedural Tasks (CVPR 2025 · similarity 0.85)
- CaptainCook4D: A Dataset for Understanding Errors in Procedural Activities (NeurIPS 2024 · similarity 0.83)
- Differentiable Task Graph Learning: Procedural Activity Representation and Online Mistake Detection from Egocentric Videos (NeurIPS 2024 · similarity 0.83)
- Gazing Into Missteps: Leveraging Eye-Gaze for Unsupervised Mistake Detection in Egocentric Videos of Skilled Human Activities (CVPR 2025 · similarity 0.83)
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.
