Object detectors trained on clear-weather data are severely impaired by adverse weather such as fog, rain, snow, and low illumination. According to the paper, these conditions introduce structured visibility degradations that the detectors are not equipped to handle.

The paper proposes a contrastive hypergraph framework for source-free domain adaptive object detection. The 'source-free' aspect means the adaptation does not rely on access to the original source dataset, which may be unavailable in practice. The method is described in an extended version of the paper posted on arXiv.

As this is a single source, there are no conflicting findings to compare. The abstract does not include experimental results or comparisons, so the effectiveness of the approach is not assessed here.