YOLOX: Exceeding YOLO Series in 2021

Benchmark Model Rank Results
object-detection-on-waterscenesYOLOX-M#2mAP@50-95: 57.8
2d-object-detection-on-ceymoYOLOX#4mAP: 57.7
real-time-object-detection-on-coco-test-devYOLOv5-X (reported by YOLOX)#17box AP: 50.4FPS (batch 1, GPU in brackets): 62.5 (V100)
object-detection-on-coco-oYOLOX-X#19Average mAP: 30.3Effective Robustness: 7.26
object-detection-on-coco-oYOLOX-S#29Average mAP: 20.6Effective Robustness: 2.48
object-detection-on-coco-test-devYOLOX-x(Modified CSP v5, 640x640, single-scale)#73box mAP: 51.5
object-detection-on-coco-test-devYOLOX-X (Modified CSP v5)#79box mAP: 51.2AP50: 69.6AP75: 55.7APS: 31.2APM: 56.1APL: 66.1…
object-detection-on-coco-test-devYOLOX-Darknet53(Darknet53, 640x640, single-scale)#102box mAP: 48.0
object-detection-on-coco-val2017YOLOX-Tiny (416)#441box AP: 32.8Params (M): 5.06
object-detection-on-coco-val2017YOLOX-Nano (416)#443box AP: 25.3Params (M): 0.91