YOLOv7: Trainable bag-of-freebies sets new state-of the art for real-time object detectorsMotivationIn the recent years the real-time object detector focus on architecture for different edge device or GPU. Authors focus on the optimized moduels and optimization methods which may increasing training cost for improving the accuracy of obejct detection, but without increasing the inference cost. Th..

OTA: Optimal Transport Assignment for Object DetectionMotivationAssigning positive training samples is technique, which boost training speed and performance. Previous almost positive training samples are assigned independently(ex. iou_threshold or best match). these ignore a fact that for objects with diffrent sizes, shapes or occlusion condition. In this paper authors assign the positive sample..