Research map / Video / Video-language understanding
GIFT: Global Irreplaceability Frame Targeting for Efficient Video Understanding
Junpeng Ma, Sashuai Zhou, Guanghao Li, Xin Gao, Yue Cao, Hengyu Zeng, Yuxiang Yan, Zhibin Wang, Jun Song, Bo Zheng, Shanghang Zhang, Jian Pu
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 GIFT frame selection instead of uniform sampling (32 frames) | LLaVA-Video-7B · 4-benchmark average | score: 60.3 -> 62.9 | Improved |
| add GIFT instead of uniform sampling (4 frames) | LLaVA-Video-7B · 4-benchmark average | score: 53.5 -> 58.7 | Improved |
| GIFT vs AKS at 32 frames | LLaVA-Video-7B · 4-benchmark average | score: 61.8 -> 62.9 | Improved |
| add GIFT to LLaVA-OneVision (8 frames) | LLaVA-OneVision-7B · VideoMME overall | accuracy: 53.9 -> 58.8 | Improved |
| add GIFT to VideoLLaMA3 (8 frames) | VideoLLaMA3-7B · MLVU | accuracy: 59.1 -> 70.7 | Improved |
| add GIFT to VideoLLaMA3 (32 frames) | VideoLLaMA3-7B · LongVideoBench | accuracy: 58.3 -> 58.9 | No clear effect |
| add GIFT to Qwen2.5-VL (8 frames) | Qwen2.5-VL-7B · LongVideoBench | accuracy: 52.7 -> 58.3 | Improved |
| use Directed Diversity instead of standard undirected diversity (ablation) | GIFT · LongVideoBench (relative to uniform) | relative performance (%): 101.7 -> 103.8 | Improved |
| add Budget-Aware Refinement (ablation) | GIFT · LongVideoBench and MLVU | relative performance: 0.5% and 2.7% gain | Improved |
| vary refinement batch size B from 6 to 12 (ablation) | GIFT · 4-benchmark average (32 frames) | average score: 62.5 (B=6), 62.6 (B=9), 61.9 (B=12) | No clear effect |
Closest papers by meaning
- BOLT: Boost Large Vision-Language Model Without Training for Long-form Video Understanding (CVPR 2025 · similarity 0.89)
- Adaptive Keyframe Sampling for Long Video Understanding (CVPR 2025 · similarity 0.88)
- Flexible Frame Selection for Efficient Video Reasoning (CVPR 2025 · similarity 0.88)
- Frame-Voyager: Learning to Query Frames for Video Large Language Models (ICLR 2025 · similarity 0.87)
- One Token per Highly Selective Frame: Towards Extreme Compression for Long Video Understanding (NeurIPS 2025 · similarity 0.87)
- PAVE: Patching and Adapting Video 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.
