Research map / Video / Video-language understanding
Sim-DETR: Unlock DETR for Temporal Sentence Grounding
Jiajin Tang, Zhengxuan Wei, Yuchen Zhu, Cheng Shi, Guanbin Li, Liang Lin, Sibei Yang
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 IoU loss (ablation) | TR-DETR baseline · QVHighlights val | mAP avg: 44.97 -> 45.22 | Improved |
| add query grouping and ranking (QGR) (ablation) | TR-DETR baseline · QVHighlights val | mAP avg: 44.97 -> 47.03 | Improved |
| add global-local bridging (GLB) (ablation) | TR-DETR baseline · QVHighlights val | mAP avg: 44.97 -> 48.17 | Improved |
| add GLB without QGR (ablation) | TR-DETR baseline · QVHighlights val | R1@0.5: -3.1% | Hurt |
| add both QGR and GLB (ablation) | TR-DETR baseline · QVHighlights val | mAP avg: 44.97 -> 49.50 | Improved |
| use span border distance instead of center distance for query relevance (ablation) | QGR · QVHighlights val | mAP avg: 48.59 -> 49.50 | Improved |
| use span border distance instead of span IoU for query relevance (ablation) | QGR · QVHighlights val | mAP avg: 48.94 -> 49.50 | Improved |
| add outer global semantics (confidence) to query relevance (ablation) | QGR · QVHighlights val | mAP avg: 48.93 -> 49.50 | Improved |
| add outer local metric (IoU prediction) to query prioritization (ablation) | QGR · QVHighlights val | mAP avg: 48.66 -> 49.50 | Improved |
| use IoU prediction instead of center prediction as local metric (ablation) | QGR · QVHighlights val | mAP avg: 48.97 -> 49.50 | Improved |
| increase number of queries to 15 and 20 | TaskWeave · QVHighlights val | performance: drop of over -2% | Hurt |
| increase number of decoder layers | CG-DETR / TaskWeave · QVHighlights val | performance: -1% to -2.5% | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- BAM-DETR — BAM-DETR: Boundary-Aligned Moment Detection Transformer for Temporal Sentence Grounding in Videos (ECCV 2024) · better on 5 of 5 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
- Decoupling Moment from Event for Video Temporal Grounding (ECCV 2026 · similarity 0.91)
- Empower Words: DualGround for Structured Phrase and Sentence-Level Temporal Grounding (NeurIPS 2025 · similarity 0.89)
- Temporal Sentence Grounding with Relevance Feedback in Videos (NeurIPS 2024 · similarity 0.88)
- Diversifying Query: Region-Guided Transformer for Temporal Sentence Grounding (AAAI 2025 · similarity 0.88)
- Multi-Pair Temporal Sentence Grounding via Multi-Thread Knowledge Transfer Network (AAAI 2025 · similarity 0.87)
- Universal Video Temporal Grounding with Generative Multi-modal Large Language Models (NeurIPS 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.
