Deformable ConvNets v2: More Deformable, Better Results

Benchmark Model Rank Results
object-detection-on-cocoDCNv2 (ResNet-101, multi-scale)#131box mAP: 46.0AP50: 67.9AP75: 50.8APS: 27.8APM: 49.1APL: 59.5
object-detection-on-coco-minivalMask R-CNN (ResNet-101, DCNv2)#145box AP: 43.1
object-detection-on-coco-minivalFaster R-CNN (ResNet-101, DCNv2)#160box AP: 41.7APS: 22.2APM: 45.8APL: 58.7