当前位置：网站首页>Swin_Unet & Trans_UNet & Unet & Deeplabv3 network training results comparison

Swin_Unet & Trans_UNet & Unet & Deeplabv3 network training results comparison

2022-08-07 11:10:00 【fat studio】

The following pictures refer to the length and width values resized by the training network. For example, 224 * 224 is the size of the training resize picture, and the final picture used is 640 * 640 size.
Insert picture description here
Overall, 224*224 Swin_Unet is the best for recognition, followed by Trans_UNet, Unet again, and Deeplabv3 last