EMYOLO: Enhanced Multispectral YOLO for Robust Object Detection in Adverse Weather

Object detection is a fundamental task in computer vision and has many important applications, such as autonomous driving. However, detecting objects in adverse weather conditions, such as rain, fog, haze, and low light, remains a difficult challenge. In order to address this issue, we propose a novel object detection framework called Enhanced Multispectral YOLO (EMYOLO). EMYOLO utilizes two key modules: the Enhanced module and the Transformer Fusing module, to enhance the representation power and feature representation of both RGB and IR modalities. We evaluate EMYOLO on several benchmark datasets, including LLVIP, TNO, and M3FD, under various weather conditions. Experimental results demonstrate that EMYOLO outperforms several object detection models and maintains a good balance between detection accuracy and model complexity, particularly in adverse weather conditions. Our proposed approach has significant implications for the development of intelligent cars and other applications that rely on object detection in challenging weather conditions.