Research map / Video / Video-language understanding
HERO: Hierarchical Embedding-Refinement for Open-Vocabulary Temporal Sentence Grounding in Videos
Tingting Han, Xinsong Tao, Yufei Yin, Min Tan, Sicheng Zhao, Zhou Yu
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 HEM, SGVF and CMTR together to the EMB baseline (ablation) | HERO (EMB-based) · Charades-OV test-ov | R1@0.5: 42.31 -> 45.51 | Improved |
| add HEM, SGVF and CMTR together to the EMB baseline (ablation) | HERO (EMB-based) · Charades-OV test-ov | R1@0.7: 25.22 -> 27.20 | Improved |
| add HEM alone (ablation) | HERO (EMB-based) · Charades-OV test-ov | R1@0.7: 25.22 -> 24.89 | Hurt |
| add SGVF alone (ablation) | HERO (EMB-based) · Charades-OV test-ov | R1@0.5: 42.31 -> 43.58 | Improved |
| add CMTR alone (ablation) | HERO (EMB-based) · Charades-OV test-ov | R1@0.5: 42.31 -> 42.24 | No clear effect |
| add SGVF on top of HEM (ablation) | HERO (EMB-based) · Charades-OV test-ov | R1@0.5: 42.64 -> 45.10 | Improved |
| add CMTR on top of HEM+SGVF (ablation) | HERO (EMB-based) · Charades-OV test-ov | R1@0.7: 26.70 -> 27.20 | Improved |
| add HEM on top of SGVF+CMTR (ablation) | HERO (EMB-based) · Charades-OV test-ov | R1@0.7: 25.21 -> 27.20 | Improved |
| add CMTR on top of HEM only (ablation) | HERO (EMB-based) · Charades-OV test-ov | R1@0.5: 42.64 -> 42.73 | Mixed |
| use four HEM levels instead of two or eight (ablation) | HERO · Charades-OV | R1 | Improved |
| replace EMB with HERO on Charades-STA (closed vocabulary) | EMB · Charades-STA | R1@0.7: 39.25 -> 41.29 | Improved |
| replace EMB with HERO on ActivityNet-OV | EMB · ActivityNet-OV test-ov | R1@0.5: 21.70 -> 25.23 | Improved |
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.
- R2-tuning — R^2-Tuning: Efficient Image-to-Video Transfer Learning for Video Temporal Grounding (ECCV 2024) · 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
- Empower Words: DualGround for Structured Phrase and Sentence-Level Temporal Grounding (NeurIPS 2025 · similarity 0.92)
- Universal Video Temporal Grounding with Generative Multi-modal Large Language Models (NeurIPS 2025 · similarity 0.90)
- Multi-Pair Temporal Sentence Grounding via Multi-Thread Knowledge Transfer Network (AAAI 2025 · similarity 0.90)
- VideoGrounding-DINO: Towards Open-Vocabulary Spatio-Temporal Video Grounding (CVPR 2024 · similarity 0.90)
- Temporal Sentence Grounding with Relevance Feedback in Videos (NeurIPS 2024 · similarity 0.89)
- ReVisionLLM: Recursive Vision-Language Model for Temporal Grounding in Hour-Long Videos (CVPR 2025 · similarity 0.89)
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.
