Reproduce “FogTTA: Online Test-Time Adaptation for Robust Transformer-based Object Detection in Foggy Weather” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ali Abedi, Q. M. Jonathan Wu, Ning Zhang, Shiqi Tian
FogTTA adapts a clear-weather RF-DETR detector to unlabeled foggy images online, using its predictions to train a student model. It filters low-confidence pseudo-labels, reweights remaining supervision with Varifocal loss, and updates the teacher through exponential moving averages. On Cityscapes to Foggy Cityscapes, it reaches 57.6% mAP versus 56.1% without adaptation, on this synthetic-fog benchmark.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.