Research map / Video / Action recognition
Comprehensive Instructional Video Analysis: The COIN Dataset and Performance Evaluation
Yansong Tang, Jiwen Lu, Jie Zhou
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 task-consistency (TC) | SSN-RGB · COIN step localization | mAP@0.5: 7.79 -> 9.33 | Improved |
| add ordering-dependency (OD) | SSN-RGB · COIN step localization | mAP@0.1: 19.39 -> 20.39 | Improved |
| apply OD then TC (ODTC) | SSN-RGB · COIN step localization | mAP@0.1: 19.39 -> 23.23 | Improved |
| apply TC then OD instead of OD then TC (ablation) | SSN-RGB · COIN step localization | mAP@0.1: 23.23 -> 21.82 | Hurt |
| sum TC scores instead of averaging (ablation) | SSN-RGB · COIN step localization | mAP@0.5: 9.16 -> 9.33 | Improved |
| add TC | R-C3D · COIN step localization | mAP@0.1: 9.85 -> 10.32 | Improved |
| add OD | BSN · COIN step localization | mAP@0.1: 18.91 -> 20.46 | Improved |
| add OD | BMN · COIN step localization | mAP@0.1: 18.60 -> 19.68 | Improved |
| add TC and OD | SSN · Breakfast step localization | mAP@0.1: 28.24 -> 30.91 | Improved |
| add TC alone | SSN · Breakfast step localization | mAP@0.1: 28.24 -> 28.25 | No clear effect |
| use Gaussian beta=1 instead of triangle distribution for OD (ablation) | SSN+OD · COIN step localization | mAP@0.1: 19.77 -> 20.39 | Improved |
| use weighted sum instead of max-pool to fuse OD scores (ablation) | SSN+OD · COIN step localization | mAP@0.1: 19.84 -> 20.30 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- R-C3D — R-C3D: Region Convolutional 3D Network for Temporal Activity Detection (ICCV 2017) · better on 1 of 1 shared result
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
- COIN: A Large-Scale Dataset for Comprehensive Instructional Video Analysis (CVPR 2019 · similarity 0.97)
- GAIN: On the Generalization of Instructional Action Understanding (ICLR 2023 · similarity 0.87)
- TL;DW? Summarizing Instructional Videos with Task Relevance & Cross-Modal Saliency (ECCV 2022 · similarity 0.87)
- Collaborative Weakly Supervised Video Correlation Learning for Procedure-Aware Instructional Video Analysis (AAAI 2024 · similarity 0.87)
- ExAct: A Video-Language Benchmark for Expert Action Analysis (NeurIPS 2025 · similarity 0.85)
- From Raw Video to Pedagogical Insights: A Unified Framework for Student Behavior Analysis (AAAI 2024 · similarity 0.85)
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.
