Contrastive Hypergraph Source-free Domain Adaptive Object Detection in Adverse Weather Conditions--Extended Version

Key points

• Adverse weather conditions such as fog, rain, snow, and low illumination introduce structured visibility degradations that severely impair object detectors trained on clear-weather data.1
• When source data are inaccessible due to privacy or trans- mission constraints, source-free domain adaptation (SFDA) becomes necessary.1
• REUNION constructs a contrastive hypergraph over target-domain object proposals, encoding high-order relations through intra-image context, weather-aware grouping, prototype-based semantic anchors, and uncertainty-aware connections between reliable and low-confidence instances.1
• To effectively exploit these structured relations, we introduce a group-wise Hyper-InfoNCE objective that optimizes representations at the hyperedge level, enabling semantic information to propagate from confident proposals to corrupted or low-contrast instances.1

Sources (1)

• [1]Contrastive Hypergraph Source-free Domain Adaptive Object Detection in Adverse Weather Conditions--Extended Version
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 17, 07:45 AM

  We propose REUNION, a hypergraph-guided SFDA framework for object detection in adverse weather conditions.

  Adverse weather conditions such as fog, rain, snow, and low illumination introduce structured visibility degradations that severely impair object detectors trained on clear-weather data.

Extractive summary: sentences quoted from the sources.