Research map / Video / Action recognition
RCL: Recurrent Continuous Localization for Temporal Action 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 |
|---|---|---|---|
| replace discretized anchoring with RCL continuous anchoring | temporal action detection · THUMOS14 test | mAP@0.5: 40.2 -> 52.9 | Helped |
| use RCL instead of VSGN | temporal action detection · ActivityNet v1.3 validation, TSP features | average mAP: 35.94 -> 37.65 | Helped |
| add CAR to the BMN baseline (ablation) | BMN · ActivityNet v1.3 validation, TSN features | average mAP: 33.85 -> 35.41 | Helped |
| add RRM to CAR (ablation) | RCL with TSN features · ActivityNet v1.3 validation | average mAP: 35.41 -> 35.98 | Helped |
| add RRM to CAR (ablation) | RCL with TSN features · ActivityNet v1.3 validation | mAP@0.75: 36.45 -> 36.19 | Hurt |
| increase the discretized grid input dimension from 100 to 200 (ablation) | BMN · ActivityNet v1.3 validation | average mAP: 33.85 -> 33.54 | Hurt |
| increase the discretized grid input dimension from 100 to 300 (ablation) | BMN · ActivityNet v1.3 validation | average mAP: 33.85 -> 33.95 | Helped |
| replace a discretized output grid with CAR and offset regression (ablation) | BMN · ActivityNet v1.3 validation | average mAP: 33.85 -> 35.41 | Helped |
| add offset regression to CAR (ablation) | RCL · ActivityNet v1.3 validation | average mAP: 34.81 -> 35.41 | Helped |
| add uniform samples to regular grid samples (ablation) | CAR · ActivityNet v1.3 validation | average mAP: 33.85 -> 34.14 | Helped |
| add scale-invariant samples (ablation) | CAR · ActivityNet v1.3 validation | average mAP: 34.14 -> 34.81 | Helped |
| use scale-invariant sampling instead of scale-invariant loss (ablation) | CAR · ActivityNet v1.3 validation | average mAP: 34.15 -> 34.81 | Helped |
Closest papers by meaning
- Rethinking the Faster R-CNN Architecture for Temporal Action Localization (CVPR 2018 · similarity 0.88)
- Learning Salient Boundary Feature for Anchor-free Temporal Action Localization (CVPR 2021 · similarity 0.88)
- CDC: Convolutional-De-Convolutional Networks for Precise Temporal Action Localization in Untrimmed Videos (CVPR 2017 · similarity 0.87)
- Bridge the Gap: From Weak to Full Supervision for Temporal Action Localization with PseudoFormer (CVPR 2025 · similarity 0.86)
- Test-Time Zero-Shot Temporal Action Localization (CVPR 2024 · similarity 0.86)
- Realigning Confidence with Temporal Saliency Information for Point-Level Weakly-Supervised Temporal Action Localization (CVPR 2024 · similarity 0.86)
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.