As explained in YOLOX paper published by Zheng GE and Team, they have switched the YOLO detector to an anchor-free manner and conduct other advanced detection techniques, i.e., a decoupled head and the leading label assignment strategy SimOTA to achieve state-of-the-art results across a large scale range of models. They have won the 1st Place on Streaming Perception Challenge (Workshop on Autonomous Driving at CVPR 2021) using a single YOLOX-L model. Megvii researchers have cleverly integrated and combined outstanding progress in the field of object detection such as decoupling, data enhancement, anchorless and label classification with YOLO, and proposed YOLOX, which not only achieves AP that surpasses YOLOv3, YOLOv4 and YOLOv5 , but also achieved a very competitive reasoning speed. As this is very recent development in YOLO Series; one may face some issues while adapting this model on their custom dataset. In this post, we will walk through how you can train YOLOX to recognize object detection on your custom image Dataset.