Research map / Video / Action recognition
Temporal Convolutional Networks for Action Segmentation and Detection
Check 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 |
|---|---|---|---|
| Use ED-TCN instead of the strongest listed baseline | 50 Salads (higher) · 50 Salads (higher) | F1@10: 72.2 -> 76.5 | Helped |
| Use ED-TCN instead of the strongest listed baseline | 50 Salads (mid) · 50 Salads (mid) | F1@10: 62.6 -> 68.0 | Helped |
| Use ED-TCN rather than MSN Seg | MERL Shopping (acausal) · MERL Shopping (acausal) | F1@10: 80.0 -> 86.7 | Helped |
| Use ED-TCN rather than MSN Det | MERL Shopping (acausal) · MERL Shopping (acausal) | mAP: 81.9 -> 74.4 | Hurt |
| Use ED-TCN instead of Dilated TCN | MERL Shopping (causal) · MERL Shopping (causal) | F1@10: 72.7 -> 82.1 | Helped |
| Use acausal rather than causal predictions | ED-TCN · MERL Shopping | F1@10: 82.1 -> 86.7 | Helped |
| Use ED-TCN instead of Bi-LSTM | GTEA · GTEA | F1@10: 66.5 -> 72.2 | Helped |
| Use ED-TCN instead of EgoNet+TDD | GTEA · GTEA | accuracy: 64.4 -> 64.0 | Hurt |
| Use NReLU rather than Sigmoid, ReLU, Tanh, or GPC (ablation) | ED-TCN · 50 Salads (mid) | F1@25: 37.3, 40.4, 48.1, 52.7 -> 58.4 | Helped |
| Use different activation functions (ablation) | Dilated TCN · 50 Salads (mid) | F1@25: 42.5, 43.1, 41.0, 40.5, 40.7 | No clear effect |
| Use longer convolutional filters in ED-TCN rather than short filters in Dilated TCN | TCN action segmentation · 50 Salads | F1@k | Helped |
| Add future-frame information | TCN action segmentation · MERL Shopping | F1@10: 82.1 -> 86.7 | Helped |
Closest papers by meaning
- MS-TCN: Multi-Stage Temporal Convolutional Network for Action Segmentation (CVPR 2019 · similarity 0.95)
- CDC: Convolutional-De-Convolutional Networks for Precise Temporal Action Localization in Untrimmed Videos (CVPR 2017 · similarity 0.92)
- T-C3D: Temporal Convolutional 3D Network for Real-Time Action Recognition (AAAI 2018 · similarity 0.91)
- Temporal Action Localization by Structured Maximal Sums (CVPR 2017 · similarity 0.90)
- Iterative Contrast-Classify for Semi-supervised Temporal Action Segmentation (AAAI 2022 · similarity 0.90)
- Weakly Supervised Action Localization by Sparse Temporal Pooling Network (CVPR 2018 · similarity 0.90)
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.