Research map / Video / Video-language understanding
TALL: Temporal Activity Localization via Language Query
Jiyang Gao, Chen Sun, Zhenheng Yang, 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 |
|---|---|---|---|
| add non-parameterized temporal regression to alignment (ablation) | CTRL · TACoS | R@1 IoU=0.5: 10.67 -> 13.30 | Improved |
| add non-parameterized temporal regression to alignment (ablation) | CTRL · Charades-STA | R@1 IoU=0.5: 18.77 -> 23.63 | Improved |
| use non-parameterized instead of parameterized offsets (ablation) | CTRL · TACoS | R@5 IoU=0.5: 23.05 -> 25.42 | Improved |
| use non-parameterized instead of parameterized offsets (ablation) | CTRL · Charades-STA | R@5 IoU=0.7: 26.61 -> 29.52 | Improved |
| add SCNN-style overlap loss to alignment (ablation) | CTRL · TACoS | R@1 IoU=0.5: 10.67 -> 10.70 | No clear effect |
| add SCNN-style overlap loss to alignment (ablation) | CTRL · Charades-STA | R@1 IoU=0.5: 18.77 -> 20.19 | Improved |
| use CTRL alignment network instead of cosine-similarity retrieval (VSA-STV) | CTRL(aln) · TACoS | R@5 IoU=0.5: 15.50 -> 19.44 | Improved |
| add object classifiers to verb classifiers (ablation) | Verb baseline · TACoS | R@1 IoU=0.5: 1.62 -> 8.25 | Improved |
| use Skip-thought instead of word2vec+LSTM sentence embedding (ablation) | CTRL(aln) · TACoS | Recall | Improved |
| use C3D instead of LRCN or VGG mean pooling visual features (ablation) | CTRL(aln) · TACoS | Recall | Improved |
| late fusion of per-sentence predictions for complex queries (ablation) | CTRL · Charades-STA complex queries | R@1 IoU=0.5: 24.09 -> 25.82 | Improved |
| apply temporal regression to proposal generation as well (ablation) | reg-np · THUMOS-14 | mAP: 19.8 -> 20.5 | Improved |
Closest papers by meaning
- Semantic Proposal for Activity Localization in Videos via Sentence Query (AAAI 2019 · similarity 0.92)
- Hierarchical Visual-Textual Graph for Temporal Activity Localization via Language (ECCV 2020 · similarity 0.91)
- Boosting Weakly-Supervised Temporal Action Localization With Text Information (CVPR 2023 · similarity 0.90)
- Language-Driven Temporal Activity Localization: A Semantic Matching Reinforcement Learning Model (CVPR 2019 · similarity 0.90)
- Divide and Conquer for Single-Frame Temporal Action Localization (ICCV 2021 · similarity 0.89)
- Weakly Supervised Temporal Action Localization via Dual-Prior Collaborative Learning Guided by Multimodal Large Language Models (CVPR 2025 · similarity 0.87)
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.
