Research map / Video / Video-language understanding
CLEVRER: Collision Events for Video Representation and Reasoning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| combine object-centric video perception, dynamics prediction, question parsing, and symbolic program execution | NS-DR · CLEVRER | descriptive accuracy: 86.4 -> 88.1 | Helped |
| combine object-centric video perception, dynamics prediction, question parsing, and symbolic program execution | NS-DR · CLEVRER explanatory | per-question accuracy: 23.7 -> 79.6 | Helped |
| combine object-centric video perception, dynamics prediction, question parsing, and symbolic program execution | NS-DR · CLEVRER predictive | per-question accuracy: 48.9 -> 68.7 | Helped |
| combine object-centric video perception, dynamics prediction, question parsing, and symbolic program execution | NS-DR · CLEVRER counterfactual | per-question accuracy: 25.1 -> 42.2 | Helped |
| add object-aware features with instance segmentation masks and intrinsic attributes | MAC (V) · CLEVRER descriptive | accuracy: 85.6 -> 86.4 | Helped |
| add object-aware features with instance segmentation masks and intrinsic attributes | MAC (V) · CLEVRER explanatory | per-option accuracy: 59.5 -> 70.5 | Helped |
| add object-aware features with instance segmentation masks and intrinsic attributes | MAC (V) · CLEVRER predictive | per-question accuracy: 16.5 -> 42.9 | Helped |
| add object-aware features with instance segmentation masks and intrinsic attributes | MAC (V) · CLEVRER counterfactual | per-question accuracy: 13.7 -> 25.1 | Helped |
| replace explicit event prediction with collision detection from object velocity changes (ablation) | NS-DR · CLEVRER | per-question accuracy across question types: 79.6 -> 74.3; 68.7 -> 54.1; 42.2 -> 42.0 | Mixed |
| train the question parser with 1,000 programs (ablation) | NS-DR · CLEVRER | accuracy across question types | Helped |
Closest papers by meaning
- Black Swan: Abductive and Defeasible Video Reasoning in Unpredictable Events (CVPR 2025 · similarity 0.90)
- When Thinking Drifts: Evidential Grounding for Robust Video Reasoning (NeurIPS 2025 · similarity 0.89)
- Look, Remember and Reason: Grounded Reasoning in Videos with Language Models (ICLR 2024 · similarity 0.89)
- Video-of-Thought: Step-by-Step Video Reasoning from Perception to Cognition (ICML 2024 · similarity 0.88)
- CLiViS: Unleashing Cognitive Map through Linguistic-Visual Synergy for Embodied Visual Reasoning (CVPR 2026 · similarity 0.88)
- SciVideoBench: Benchmarking Scientific Video Reasoning in Large Multimodal Models (ICML 2026 · 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.