Research map / Video / Action recognition
Error Recognition in Procedural Videos using Generalized Task Graph
Shih-Po Lee, Ehsan Elhamifar
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 |
|---|---|---|---|
| replace majority voting with the Error Recognition Module (ERM) (ablation) | GTG2Vid · EgoPER Quesadilla | w-F1@0: 5.1 -> 31.7 | Improved |
| replace majority voting with ERM (ablation) | GTG2Vid · EgoPER Oatmeal | w-F1@0: 2.7 -> 31.3 | Improved |
| add normal type features to the joint similarity score (ablation) | GTG2Vid · EgoPER Quesadilla | w-F1@0: 16.3 -> 31.7 | Improved |
| add normal type features to the joint similarity score (ablation) | GTG2Vid · EgoPER Pinwheel | w-F1@0: 29.9 -> 17.8 | Hurt |
| use ERM joint similarity instead of cosine similarity (ablation) | ERM · EgoPER Pinwheel | Top-1 accuracy: 27.7 -> 38.4 | Improved |
| use ERM instead of cosine similarity (ablation) | ERM · EgoPER Quesadilla | Top-1 accuracy: 42.6 -> 54.6 | Improved |
| use ERM instead of cosine similarity (ablation) | ERM · EgoPER Tea | Top-2 accuracy: 45.3 -> 53.7 | Improved |
| use dynamic frame dropping cost instead of naive percentile cost of G2Vid (ablation) | GTG2Vid · EgoPER Pinwheel | F1@.5: 23.2 -> 26.0 | Improved |
| use dynamic frame dropping cost instead of naive percentile cost of G2Vid (ablation) | GTG2Vid · EgoPER Quesadilla | F1@.5: 30.9 -> 33.8 | Improved |
| use GTG2Vid instead of EgoPED for TAS | GTG2Vid · EgoPER All | F1@.5: 56.4 -> 69.6 | Improved |
| use GTG2Vid instead of EgoPED for error detection | GTG2Vid · EgoPER All | ED F1@.5: 22.5 -> 33.0 | Improved |
| use GTG2Vid instead of G2Vid on edit score | GTG2Vid · EgoPER All | TAS Edit: 70.0 -> 73.9 | No clear effect |
Closest papers by meaning
- Error Detection in Egocentric Procedural Task Videos (CVPR 2024 · similarity 0.89)
- Differentiable Task Graph Learning: Procedural Activity Representation and Online Mistake Detection from Egocentric Videos (NeurIPS 2024 · similarity 0.88)
- 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 0.84)
- PREGO: Online Mistake Detection in PRocedural EGOcentric Videos (CVPR 2024 · similarity 0.84)
- PHGC: Procedural Heterogeneous Graph Completion for Natural Language Task Verification in Egocentric Videos (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.
