Research map / Video / Video-language understanding
Moment Quantization for Video Temporal Grounding
Xiaolong Sun, Le Wang, Sanping Zhou, Liushuai Shi, Kun Xia, Mengnan Liu, Yabing Wang, Gang Hua
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 |
|---|---|---|---|
| quantize after temporal modeling instead of no quantization (ablation) | MQVTG baseline · QVHighlights val | MR mAP Avg: 45.63 -> 47.02 | Improved |
| add soft quantization with continuous features (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 47.02 -> 47.54 | Improved |
| add moment codebook (prior init and joint projection) (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 47.54 -> 48.81 | Improved |
| moment quantization instead of image quantization (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 46.55 -> 48.81 | Improved |
| moment quantization instead of clip quantization (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 46.93 -> 48.81 | Improved |
| soft quantization instead of hard discrete features (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 47.46 -> 48.81 | Improved |
| soft quantization instead of concatenating discrete features (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 47.60 -> 48.81 | Improved |
| soft quantization instead of adding discrete features (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 47.88 -> 48.81 | Improved |
| k-means clustering init instead of random init (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 46.89 -> 48.81 | Improved |
| k-means init instead of random selection of clip features (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 47.59 -> 48.81 | Improved |
| learnable projected codebook instead of frozen codebook (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 46.92 -> 48.81 | Improved |
| joint projector instead of basic learnable codebook (ablation) | MQVTG · QVHighlights val | MR mAP Avg: 47.86 -> 48.81 | 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 5 of 7 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
- When One Moment Isn't Enough: Multi-Moment Retrieval with Cross-Moment Interactions (NeurIPS 2025 · similarity 0.89)
- Universal Video Temporal Grounding with Generative Multi-modal Large Language Models (NeurIPS 2025 · similarity 0.89)
- Empower Words: DualGround for Structured Phrase and Sentence-Level Temporal Grounding (NeurIPS 2025 · similarity 0.89)
- Multi-Scale Contrastive Learning for Video Temporal Grounding (AAAI 2025 · similarity 0.89)
- Transferable Video Moment Localization by Moment-Guided Query Prompting (AAAI 2024 · similarity 0.88)
- Time-R1: Post-Training Large Vision Language Model for Temporal Video Grounding (NeurIPS 2025 · similarity 0.88)
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.
