Research map / Video / Action recognition
Iterative Contrast-Classify for Semi-supervised Temporal Action Segmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Apply ICC with all training videos labeled instead of supervised C2F-TCN | C2F-TCN · Breakfast | MoF: 73.4 -> 75.2 | Helped |
| Apply ICC with all training videos labeled instead of supervised C2F-TCN | C2F-TCN · 50Salads | MoF: 79.4 -> 85.0 | Helped |
| Apply ICC with all training videos labeled instead of supervised C2F-TCN | C2F-TCN · GTEA | MoF: 79.5 -> 82.0 | Helped |
| Add time-proximity condition to cluster-based contrastive learning (ablation) | Unsupervised representation learning · Breakfast | MoF: 36.1 -> 50.0 | Helped |
| Add time-proximity condition to cluster-based contrastive learning (ablation) | Unsupervised representation learning · 50Salads | MoF: 50.8 -> 51.6 | No clear effect |
| Add video-level contrastive loss (ablation) | Unsupervised representation learning · Breakfast | MoF: 50.0 -> 66.1 | Helped |
| Combine normalized decoder features at multiple temporal resolutions rather than use only z6 (ablation) | Unsupervised representation learning · 50Salads | MoF: 51.6 -> 62.5 | Helped |
| Remove initial unsupervised representation learning (ablation) | ICC on 50Salads with 5% labels · 50Salads | F1@10: 52.9 -> 42.6 | Hurt |
| Use 120 rather than 60 samples per video (ablation) | Unsupervised representation learning · 50Salads | F1@10: 38.5 -> 40.8 | Helped |
| Increase to 180 samples per video (ablation) | Unsupervised representation learning · 50Salads | MoF: 62.5 -> 62.1 | Hurt |
| Use approximately 2A rather than A K-means clusters (ablation) | Unsupervised representation learning · GTEA | MoF: 65.6 -> 69.1 | Helped |
| Add multi-resolution feature augmentation during training (ablation) | Unsupervised representation learning · Breakfast | MoF: 69.4 -> 70.5 | Helped |
Closest papers by meaning
- Semi-Supervised Action Recognition With Temporal Contrastive Learning (CVPR 2021 · similarity 0.92)
- SCT: Set Constrained Temporal Transformer for Set Supervised Action Segmentation (CVPR 2020 · similarity 0.91)
- Temporally-Weighted Hierarchical Clustering for Unsupervised Action Segmentation (CVPR 2021 · similarity 0.91)
- FACT: Frame-Action Cross-Attention Temporal Modeling for Efficient Action Segmentation (CVPR 2024 · similarity 0.90)
- Unsupervised Action Segmentation via Fast Learning of Semantically Consistent Actoms (AAAI 2024 · similarity 0.89)
- Hierarchical Vector Quantization for Unsupervised Action Segmentation (AAAI 2025 · similarity 0.89)
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.