YOLOv4: Optimal Speed and Accuracy of Object Detection

Benchmark Model Rank Results
object-detection-on-pku-ddd17-carYOLOv4#12mAP50: 81.3
object-detection-on-coco-oYOLOv4-P6#18Average mAP: 30.4Effective Robustness: 5.89
real-time-object-detection-on-coco-test-devYOLOv4 (CSPDarknet-53, 608)#19box AP: 43.5FPS (batch 1, GPU in brackets): 62 (V100)
real-time-object-detection-on-coco-test-devYOLOv4 (CSPDarknet-53, 512)#20box AP: 43.0FPS (batch 1, GPU in brackets): 83 (V100)
real-time-object-detection-on-coco-test-devYOLOv4 (CSPDarknet-53, 416)#21box AP: 41.2FPS (batch 1, GPU in brackets): 96 (V100)
object-detection-on-coco-test-devYOLOv4-608#142box mAP: 43.5AP50: 65.7AP75: 47.3APS: 26.7APM: 46.7APL: 53.3