Research map / Video / Action recognition
CTAP: Complementary Temporal Action Proposal Generation
Jiyang Gao, Kan Chen, Ram Nevatia
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 |
|---|---|---|---|
| replace TURN mean-pooling ranking/adjustment with temporal-convolution TAR | sliding-window proposals · THUMOS-14 test (Flow-16) | AR@200: 42.96 -> 45.08 | Improved |
| replace TURN with TAR | sliding-window proposals · ActivityNet v1.3 val | AR@100: 49.73 -> 68.02 | Improved |
| replace TURN with TAR | sliding-window proposals · ActivityNet v1.3 val | AUC: 54.16 -> 61.02 | Improved |
| replace TURN with TAR | actionness (TAG) proposals · ActivityNet v1.3 val | AUC: 53.92 -> 64.62 | Improved |
| replace TURN with TAR | actionness (TAG) proposals · ActivityNet v1.3 val | AR@100: 63.46 -> 64.01 | Improved |
| use PATE complementary filtering instead of plain union (ablation) | CTAP · THUMOS-14 test (Flow-16) | AR@50: 25.80 -> 31.03 | Improved |
| use PATE instead of union plus NMS (ablation) | CTAP · THUMOS-14 test (Flow-16) | AR@100: 39.71 -> 40.23 | Improved |
| use PATE instead of tIoU-selection (ablation) | CTAP · THUMOS-14 test (Flow-16) | AR@200: 42.41 -> 50.13 | Improved |
| add NMS (0.7) to the union of proposals (ablation) | union baseline · THUMOS-14 test (Flow-16) | AR@100: 34.70 -> 39.71 | Improved |
| add complementary sliding windows to actionness proposals (full CTAP vs TAG-TAR) (ablation) | TAG-TAR · ActivityNet v1.3 val | AR@100: 64.01 -> 73.17 | Improved |
| add actionness proposals to sliding windows via PATE (CTAP vs SW-TAR) (ablation) | SW-TAR · ActivityNet v1.3 val | AUC: 61.02 -> 65.72 | Improved |
| use CTAP proposals instead of TURN with SCNN detector | SCNN · THUMOS-14 test detection | mAP@0.5: 25.6 -> 29.9 | Improved |
Later papers that compared with this method
3 later accepted papers report CTAP as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Boundary Content Graph Neural Network for Temporal Action Proposal Generation (ECCV 2020 · better on 2 of 2 shared results)
- BMN: Boundary-Matching Network for Temporal Action Proposal Generation (ICCV 2019 · better on 2 of 2 shared results)
- Learning Temporal Action Proposals With Fewer Labels (ICCV 2019 · better on 2 of 2 shared results)
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
- Relaxed Transformer Decoders for Direct Action Proposal Generation (ICCV 2021 · similarity 0.88)
- Temporal Action Proposal Generation with Background Constraint (AAAI 2022 · similarity 0.87)
- Learning Temporal Action Proposals With Fewer Labels (ICCV 2019 · similarity 0.87)
- HyrCap: Hybrid Rank-Calibration of Action Proposals for Temporal Event Understanding (NeurIPS 2026 · similarity 0.85)
- Coherent Temporal Synthesis for Incremental Action Segmentation (CVPR 2024 · similarity 0.85)
- OnlineTAS: An Online Baseline for Temporal Action Segmentation (NeurIPS 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.
