AID: Pushing the Performance Boundary of Human Pose Estimation with Information Dropping Augmentation

Benchmark Model Rank Results
multi-person-pose-estimation-on-coco-minivalHRNet-W48plus#1AP: 79.1
multi-person-pose-estimation-on-coco-minivalHRNet-W32#2AP: 77.8
multi-person-pose-estimation-on-coco-minivalResNet50#3AP: 75.3
multi-person-pose-estimation-on-coco-test-devHRNet-W48plus#1AP: 78.7
multi-person-pose-estimation-on-coco-test-devHRNet-W32#2AP: 76.2
multi-person-pose-estimation-on-coco-test-devResNet50#3AP: 73.7