COCO

The model was evaluated on the 5K validation set from COCO 2017

ModelFull Precision mAPQuantized mAPInput Resolution (HxWxC)Params (M)Flops (G)DownloadSource
yolact_mobilenet_v1_fpn18.8318.3512x512x319.1151.92pretrained/compiledlink
yolact_regnetx_800mf_fpn*^22.6822.54512x512x321.9751.47pretrained/compiledlink

* Inference and evaluation was made on 20 indoor classes. The model performance on the full classes list is 62.14/64 mAP (bbox/segmentation)

^ networks available in TAPPAS

D2S

The model was evaluated on MVTec D2S
ModelFull Precision mAPQuantized mAPInput Resolution(HxWxC)Params (M)Flops (G)DownloadSource
yolact_regnetx_600mf_fpn*61.7560.66512x512x322.1451.62pretrained/compiledlink

* Inference and evaluation was made on 30 classes out of 60 classes found in D2s. The model performance on the full classes list is 62.14/64 mAP (bbox/segmentation)