Explicit Regional Representations For Video Quality Assessment

arXiv:2610.36978v1 Announce Type: new Abstract: End-to-end video quality assessment (VQA) methods leave the regional evidence underlying their predictions largely implicit. We introduce an explicit pixel-level regional representation for VQA, instantiated as a Quality-Relevance Heatmap (QRH) that combines localized degradation evidence with perceptual priors under vision-language model guidance, without requiring pixel-level annotations. We apply QRH to existing VQA models through two deterministic fragment-sampling strategies: distribution-preserving sampling approximates the original sampler's QRH-score distribution without retraining, while quality-aware sampling prioritizes regions with high QRH scores and is paired with fine-tuning. Experiments with DOVER on multiple benchmarks show that the former closely approximates predictions averaged over random samples, while the latter improves most evaluated metrics across popular VQA benchmarks. These results support explicit regional representations as a useful interface for downstream VQA.

Related

• DF-CBM: Region-Aware Concept Bottleneck Models for Deepfake Detection
• MorphoSHAP: Rethinking the Unit of Attribution in Explanation for Deep Visual Models
• Location-Aware Fine-Grained Representation Learning for Medical Vision Foundation Models
• TRACE: Evidence Grounding-Guided Multi-Video Event Understanding and Claim Generation

Source: arXiv cs.CV | 2026-10-28