Research map / Video / Video-language understanding
FineVAU: A Novel Human-Aligned Benchmark for Fine-Grained Video Anomaly Understanding
Joao Alexandre Cardeira Pereira, Vasco Lopes, João C. Neves, David Semedo
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 FV-Score instead of n-gram or LLM-judge metrics | VAU evaluation · Human correlation (60 UCF-Crime videos) | Pearson correlation: 0.53 -> 0.61 (vs VAU-EVAL) | Improved |
| weight Who dimension 2x in FV-Score (ablation) | FV-Score · Human correlation (60 UCF-Crime videos) | Pearson correlation: 0.51 -> 0.61 | Improved |
| weight Where dimension 2x in FV-Score (ablation) | FV-Score · Human correlation (60 UCF-Crime videos) | Pearson correlation: 0.51 -> 0.47 | Hurt |
| weight What dimension 2x in FV-Score (ablation) | FV-Score · Human correlation (60 UCF-Crime videos) | Pearson correlation: 0.51 -> 0.56 | Improved |
| use LLM-judge metrics (AnomEVAL, VAU-EVAL) instead of n-gram metrics | VAU evaluation · Human correlation (60 UCF-Crime videos) | Pearson correlation: ROUGE-L 0.47 vs AnomEVAL 0.42, VAU-EVAL 0.53 | No clear effect |
| use BLEU or CIDEr as evaluation metric | VAU evaluation · Human correlation (60 UCF-Crime videos) | Pearson correlation: CIDEr -0.63, BLEU 0.19 | Hurt |
| compare LVLM dimensions: location vs events | five open LVLMs · FineVAU | accuracy (%): Location mean 61.3 vs Event mean 12.2 | Mixed |
Closest papers by meaning
- Holmes-VAU: Towards Long-term Video Anomaly Understanding at Any Granularity (CVPR 2025 · similarity 0.92)
- HAWK: Learning to Understand Open-World Video Anomalies (NeurIPS 2024 · similarity 0.88)
- VERA: Explainable Video Anomaly Detection via Verbalized Learning of Vision-Language Models (CVPR 2025 · similarity 0.88)
- Open-Vocabulary Video Anomaly Detection (CVPR 2024 · similarity 0.88)
- MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video Understanding (NeurIPS 2024 · similarity 0.88)
- Vad-R1: Towards Video Anomaly Reasoning via Perception-to-Cognition Chain-of-Thought (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.
