List your company
ResearchFormerly WizWand

Instance Segmentation on COCO 2017 v1.0 (val)

51.2Mask APSwinV2-L
44.02445.88747.7549.613Mar 23, 2023Apr 19, 2023May 16, 2023Jun 12, 2023Jul 9, 2023Aug 5, 2023Sep 2, 2023
Updated 8mo ago

Evaluation results

MethodLinks
SwinV2-LPre-training Dataset=IN21k, Architecture=SwinV2-L, Framework=HTC++2023.0351.2
CBNetv2Pre-training Dataset=IN21k, Architecture=2 xSwin-L, Framework=HTC2023.0351
MAEPre-training Dataset=IN1k, Architecture=ViTDet-H, Framework=Cascade2023.0350.9
MAWSPre-training Dataset=IG-3B, Architecture=ViTDet-2B, Framework=Cascade2023.0350.1
MAWSPre-training Dataset=IG-3B, Architecture=ViTDet-H, Framework=Cascade2023.0349.6
RevColV2-BFramework=Cascade Mask R-CNN, Params=156M2023.0947.9
SWAGPre-training Dataset=IG-3.6B, Architecture=ViTDet-H, Framework=Cascade2023.0347.9
RevColV2-LFramework=Mask R-CNN, Params=363M2023.0947.8
MAE-LFramework=Mask R-CNN, Backbone=ViT-L, Architecture Adapter=ViT-Adapter, Params=330M2023.0947.4
MAE-LFramework=Mask R-CNN, Backbone=ViT-L, Params=323M2023.0947.1
EVA-02-B++Framework=Cascade Mask R-CNN, Pre-training=ImageNet-22K, Distillation Teacher=true, Params=141M2023.0947.1
ViTDet-BFramework=Cascade Mask R-CNN, Params=141M2023.0946.7
RevColV2-BFramework=Mask R-CNN, Params=119M2023.0946.2
RevColFramework=Cascade Mask R-CNN, Params=196M2023.0945.9
ConvNeXt-BFramework=Cascade Mask R-CNN, Params=146M2023.0945.7
MAE-BFramework=Mask R-CNN, Backbone=ViT-B, Architecture Adapter=ViT-Adapter, Params=122M2023.0945.4
Swin-BFramework=Cascade Mask R-CNN, Params=145M2023.0945
MAE-BFramework=Mask R-CNN, Backbone=ViT-B, Params=110M2023.0944.3