Research map / Video / Action recognition
Are you Struggling? Dataset and Baselines for Struggle Determination in Assembly Videos
Shijia Feng, Michael Wray, Brian Sullivan, Youngkyoon Jang, Casimir J. H. Ludwig, Iain Donald Gilchrist, Walterio W. Mayol-Cuevas
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 |
|---|---|---|---|
| train as classification instead of regression (ablation) | MViTv2 · Pipes-Struggle binary | accuracy: 75.03 -> 76.68 | Improved |
| train as classification instead of regression (ablation) | MViTv2 · Tower-Struggle four-way | accuracy: 33.41 -> 57.42 | Improved |
| train on combined struggle subsets instead of Pipes alone (ablation) | MViTv2 · Pipes-Struggle binary | accuracy: 76.68 -> 79.83 | Improved |
| train on combined struggle subsets instead of Tent alone (ablation) | MViTv2 · Tent-Struggle binary | accuracy: 64.42 -> 69.48 | Improved |
| train on combined struggle subsets instead of Tower alone (ablation) | MViTv2 · Tower-Struggle four-way | accuracy: 57.42 -> 68.81 | Improved |
| zero-shot transfer from Pipes to Tent (ablation) | MViTv2 · Tent-Struggle binary | accuracy: 64.42 -> 49.22 | Hurt |
| evaluate with a single frame instead of 128 frames (ablation) | SlowFast models · all three activities | accuracy | Hurt |
| shuffle frame order at test time (ablation) | SlowFast models · all three activities | accuracy | Hurt |
| add ViT or gMLP temporal layer on SlowFast-R50 | SlowFast-R50 · Pipes-Struggle binary | accuracy: 78.01 -> 77.80 | No clear effect |
| use SlowFast-R50 instead of TSN (2D to 3D ConvNet) | TSN · Tower-Struggle binary | accuracy: 82.59 -> 88.24 | Improved |
| use label distribution learning instead of classification | MViTv2 · Pipes-Struggle binary | accuracy: 76.68 -> 78.88 | Improved |
| use label distribution learning instead of classification | MViTv2 · Tent-Struggle binary | accuracy: 64.42 -> 64.58 | Mixed |
Closest papers by meaning
- Comprehensive Instructional Video Analysis: The COIN Dataset and Performance Evaluation (TPAMI 2020 · similarity 0.85)
- About Time: Advances, Challenges, and Outlooks of Action Understanding (IJCV 2025 · similarity 0.84)
- Assembly101: A Large-Scale Multi-View Video Dataset for Understanding Procedural Activities (CVPR 2022 · similarity 0.84)
- COIN: A Large-Scale Dataset for Comprehensive Instructional Video Analysis (CVPR 2019 · similarity 0.83)
- Group-Aware Contrastive Regression for Action Quality Assessment (ICCV 2021 · similarity 0.83)
- GAIN: On the Generalization of Instructional Action Understanding (ICLR 2023 · 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.
