Research map / Video / Video-language understanding
Grounded Video Situation Recognition
Zeeshan Khan, C.V. Jawahar, Makarand Tapaswi
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 |
|---|---|---|---|
| use role-object decoder instead of role-video decoder (ablation) | VideoWhisperer · VidSitu SRL | CIDEr: 47.91 -> 70.48 | Improved |
| add video-object encoder before role-object decoder (ablation) | VideoWhisperer · VidSitu grounding | IoU@0.3: 0.14 -> 0.29 | Improved |
| add video-object encoder before role-object decoder (ablation) | VideoWhisperer · VidSitu grounding | IoU@0.5: 0.06 -> 0.12 | Improved |
| add video-object encoder before role-object decoder (ablation) | VideoWhisperer · VidSitu SRL | CIDEr: 70.48 -> 68.54 | No clear effect |
| contextualize objects with events (VO instead of V) (ablation) | VideoWhisperer · VidSitu grounding | IoU@0.3: 0.23 -> 0.29 | Improved |
| use joint VOR encoder instead of staged RO decoder (ablation) | VideoWhisperer · VidSitu SRL | CIDEr: 70.48 -> 67.4 | Hurt |
| use event embedding instead of GT-verb in role query (ablation) | VideoWhisperer · VidSitu grounding | IoU@0.3: 0.25 -> 0.29 | Improved |
| use event embedding instead of GT-verb in role query (ablation) | VideoWhisperer · VidSitu SRL | CIDEr: 68.71 -> 68.54 | No clear effect |
| role embedding alone as query (ablation) | VideoWhisperer · VidSitu SRL | CIDEr: 68.61 -> 68.54 | No clear effect |
| add masking in RO decoder cross-attention (ablation) | VideoWhisperer · VidSitu SRL | CIDEr: 67.02 -> 68.54 | Improved |
| add masking in RO decoder cross-attention (ablation) | VideoWhisperer · VidSitu grounding | IoU@0.3: 0.25 -> 0.29 | Improved |
| replace independent clip prediction with video-only transformer encoder (ablation) | verb predictor · VidSitu | Acc@1: 46.79 -> 48.82 | Improved |
Closest papers by meaning
- Visual Semantic Role Labeling for Video Understanding (CVPR 2021 · similarity 0.92)
- Grounded Situation Recognition (ECCV 2020 · similarity 0.90)
- Comprehensive Visual Grounding for Video Description (AAAI 2024 · similarity 0.90)
- Dense Video Object Captioning from Disjoint Supervision (ICLR 2025 · similarity 0.88)
- VideoGEM: Training-free Action Grounding in Videos (CVPR 2025 · similarity 0.88)
- SAMA: Towards Multi-Turn Referential Grounded Video Chat with Large Language Models (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.
