List your company
ResearchFormerly WizWand

Image Classification on ImageNet-1K (ImageNet-21K Pre-training) (test)

87.5Top-1 AccCSWin-L
83.8684.80585.7586.695Jul 1, 2021
Updated 8mo ago

Evaluation results

MethodLinks
CSWin-LParam=173M, Size=384x384, FLOPs=96.8G, Pre-training=ImageNet-21K2021.0787.5
Swin-LParam=197M, Size=384x384, FLOPs=103.9G, Pre-training=ImageNet-21K2021.0787.3
CSWin-BParam=78M, Size=384x384, FLOPs=47.0G, Pre-training=ImageNet-21K2021.0787
CSWin-LParam=173M, Size=224x224, FLOPs=31.5G, Pre-training=ImageNet-21K2021.0786.5
Swin-BParam=88M, Size=384x384, FLOPs=47.1G, Pre-training=ImageNet-21K2021.0786.4
Swin-LParam=197M, Size=224x224, FLOPs=34.5G, Pre-training=ImageNet-21K2021.0786.3
CSWin-BParam=78M, Size=224x224, FLOPs=15.0G, Pre-training=ImageNet-21K2021.0785.9
R-152x4Param=937M, Size=480x480, FLOPs=840.5G, Pre-training=ImageNet-21K2021.0785.4
ViT-L/16Param=307M, Size=384x384, FLOPs=190.7G, Pre-training=ImageNet-21K2021.0785.2
Swin-BParam=88M, Size=224x224, FLOPs=15.4G, Pre-training=ImageNet-21K2021.0785.2
R-101x3Param=388M, Size=384x384, FLOPs=204.6G, Pre-training=ImageNet-21K2021.0784.4
ViT-B/16Param=86M, Size=384x384, FLOPs=55.4G, Pre-training=ImageNet-21K2021.0784