Research map / Video / Video-language understanding
Learning 2D Temporal Adjacent Networks for Moment Localization with Natural Language
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 |
|---|---|---|---|
| Model temporal context between moment candidates on a 2D start-end map | 2D-TAN versus MAN · Charades-STA | Rank1@0.7: 20.54 -> 23.31 | Helped |
| Model temporal context between moment candidates on a 2D start-end map | 2D-TAN versus CMIN · ActivityNet Captions | Rank1@0.7: 23.88 -> 26.54 | Helped |
| Model temporal context between moment candidates on a 2D start-end map | 2D-TAN versus ACL-K · TACoS | Rank1@0.5: 20.01 -> 25.32 | Helped |
| Increase sampled clips from N=16 to N=64 (ablation) | 2D-TAN Enum · ActivityNet Captions | Rank5@0.7: 57.79 v.s 59.66 v.s 60.18 | Helped |
| Enlarge the receptive field from kernel 1, layer 1 to kernel 5, layer 4 (ablation) | 2D-TAN Enum · ActivityNet Captions | Rank5@0.7: 37.13 -> 57.46 | Helped |
| Further enlarge the receptive field from kernel 5, layer 4 to kernel 9, layer 4 (ablation) | 2D-TAN Enum · ActivityNet Captions | Rank5@0.7: 57.46 -> 60.18 | Helped |
| Use sparse sampling instead of dense enumeration (ablation) | 2D-TAN Pool · ActivityNet Captions | Rank5@0.7: 60.18 -> 61.96 | No clear effect |
| Replace max-pooling moment features with stacked convolution (ablation) | 2D-TAN · ActivityNet Captions | Rank5@0.7: 61.96 -> 62.26 | Helped |
| Replace max-pooling moment features with stacked convolution (ablation) | 2D-TAN · Charades-STA | Rank1@0.5: 39.70 -> 39.81 | No clear effect |
| Replace max-pooling moment features with stacked convolution (ablation) | 2D-TAN · TACoS | Rank1@0.5: 25.32 -> 25.19 | No clear effect |
Closest papers by meaning
- Transferable Video Moment Localization by Moment-Guided Query Prompting (AAAI 2024 · similarity 0.90)
- Structured Multi-Level Interaction Network for Video Moment Localization via Language Query (CVPR 2021 · similarity 0.90)
- Boundary-Aware Temporal Dynamic Pseudo-Supervision Pairs Generation for Zero-Shot Natural Language Video Localization (AAAI 2025 · similarity 0.86)
- Anchor-Aware Similarity Cohesion in Target Frames Enables Predicting Temporal Moment Boundaries in 2D (CVPR 2025 · similarity 0.86)
- Universal Video Temporal Grounding with Generative Multi-modal Large Language Models (NeurIPS 2025 · similarity 0.86)
- Unsupervised Domain Adaptative Temporal Sentence Localization with Mutual Information Maximization (AAAI 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.