Research map / Video / Video-language understanding
ViSiL: Fine-Grained Spatio-Temporal Video Similarity Learning
Giorgos Kordopatis-Zilos, Symeon Papadopoulos, Ioannis Patras, Ioannis Kompatsiaris
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 whitening (ablation) | ViSiL_f · FIVR-5K DSVR | mAP: 0.838 -> 0.844 | Improved |
| add whitening (ablation) | ViSiL_f · FIVR-5K ISVR | mAP: 0.739 -> 0.750 | Improved |
| add attention mechanism (ablation) | ViSiL_f+W · FIVR-5K ISVR | mAP: 0.750 -> 0.768 | Improved |
| add attention mechanism (ablation) | ViSiL_f+W · FIVR-5K DSVR | mAP: 0.844 -> 0.856 | Improved |
| add video-level CNN for refined similarity (ablation) | ViSiL_f+W+A · FIVR-5K DSVR | mAP: 0.856 -> 0.880 | Improved |
| add video-level CNN for refined similarity (ablation) | ViSiL_f+W+A · FIVR-5K CSVR | mAP: 0.848 -> 0.869 | Improved |
| add similarity regularization loss (ablation) | ViSiL_v · FIVR-5K DSVR | mAP: 0.859 -> 0.880 | Improved |
| add similarity regularization loss (ablation) | ViSiL_v · FIVR-5K ISVR | mAP: 0.756 -> 0.777 | Improved |
| use asymmetric Chamfer similarity instead of symmetric (ablation) | ViSiL · FIVR-5K DSVR | mAP: 0.830 -> 0.880 | Improved |
| use multi-level regions (L3-iMAC) instead of iMAC (ablation) | frame features · FIVR-5K DSVR | mAP: 0.755 -> 0.838 | Improved |
| use max-pooling in frame-to-frame similarity instead of average pooling (ablation) | ViSiL_v · FIVR-5K DSVR | mAP: 0.769 -> 0.880 | Improved |
| use Chamfer (max then average) in video-to-video similarity instead of two average poolings (ablation) | ViSiL_v · FIVR-5K DSVR | mAP: 0.640 -> 0.880 | Improved |
Closest papers by meaning
- From Play to Replay: Composed Video Retrieval for Temporally Fine-Grained Videos (NeurIPS 2025 · similarity 0.86)
- ConViS-Bench: Estimating Video Similarity Through Semantic Concepts (NeurIPS 2025 · similarity 0.86)
- Mitigating Semantic Collapse in Partially Relevant Video Retrieval (NeurIPS 2025 · similarity 0.86)
- TempMe: Video Temporal Token Merging for Efficient Text-Video Retrieval (ICLR 2025 · similarity 0.85)
- Bridging the Semantic Granularity Gap Between Text and Frame Representations for Partially Relevant Video Retrieval (AAAI 2025 · similarity 0.85)
- Video-ColBERT: Contextualized Late Interaction for Text-to-Video Retrieval (CVPR 2025 · 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.
