{"id":19,"date":"2025-10-03T08:54:37","date_gmt":"2025-10-03T08:54:37","guid":{"rendered":"https:\/\/vap.aau.dk\/rws\/?page_id=19"},"modified":"2025-10-15T11:24:35","modified_gmt":"2025-10-15T11:24:35","slug":"challenge","status":"publish","type":"page","link":"https:\/\/vap.aau.dk\/rws\/challenge\/","title":{"rendered":"Challenge"},"content":{"rendered":"\n
\"\"<\/figure>\n\n\n\n
\n\n\n\n

Ready to stress-test your detector in the wild? The 6th RWS workshop introduce a new challenge on Robust Thermal-Image Object Detection to advance multi-object detection performance under long-term thermal drift.<\/strong><\/p>\n\n\n\n

Real-world thermal imaging is messy: sensors re-calibrate, ambient temperatures swing, and weather keeps changing, slowly eroding detector performance. The RWS Robust Thermal-Image Object Detection Challenge<\/strong> puts this problem front using the LTDv2<\/a><\/strong>, a large-scale dataset purpose-built for bechmarking long-term robustness wtih 1 million frames<\/strong> from video over 8 months<\/strong>, 6.8+ million annotated boxes<\/strong>, and rich weather metadata<\/strong>.

Your mission<\/strong>: build object detectors that don\u2019t just perform well on a sunny day, but stay consistent across seasons, weather patterns, and day\u2013night cycles<\/strong>. We\u2019ll score submissions on overall mAP<\/strong> and temporal consistency<\/strong> to reward models that are both accurate and stable over time. The challenge features a development phase<\/strong> for method iteration and a final test phase<\/strong> for the leaderboard. A starter kit with dataloaders and YOLOv8 baselines<\/strong> are available to get you up and running quickly. Now it\u2019s your turn to beat them<\/strong>.<\/p>\n\n\n\n

The CodaBench Robust Thermal Image Object Detection Challenge Platform can be found here: https:\/\/www.codabench.org\/competitions\/10954\/<\/a><\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

<\/h1>\n\n\n\n
\n\n\n\n

Challenge Schedule:<\/h2>\n\n\n\n