Asymmetric Loss For Multi-Label Classification

Benchmark Model Rank Results
multi-label-classification-on-ms-cocoTResNet-XL (resolution 640)#12mAP: 88.4
multi-label-classification-on-ms-cocoTResNet-L (resolution 448)#17mAP: 86.6
multi-label-classification-on-nus-wideTResNet-L (resolution 448)#4MAP: 65.2
multi-label-classification-on-openimages-v6TResNet-L#4mAP: 86.3
multi-label-classification-on-pascal-voc-2007TResNet-L (resolution 448, pretrain from MS-COCO)#8mAP: 95.8
multi-label-classification-on-pascal-voc-2007TResNet-L (resolution 448, pretrain from ImageNet)#12mAP: 94.6