TransUNet
TransUNet copied to clipboard
想问一下各位大佬,为什么我用这个框架在Synapse数据上UNet的dice能跑到80多?
本来想测试一下baseline,但是发现Unet量化结果奇高啊
本来想测试一下baseline,但是发现Unet量化结果奇高啊
是的,Unet就是最高的
那还怎么魔改啊😂
获取Outlook for Androidhttps://aka.ms/AAb9ysg
From: DiamondH @.> Sent: Friday, December 30, 2022 1:58:45 AM To: Beckschen/TransUNet @.> Cc: dayang_wang @.>; Author @.> Subject: Re: [Beckschen/TransUNet] 想问一下各位大佬,为什么我用这个框架在Synapse数据上UNet的dice能跑到80多? (Issue #110)
本来想测试一下baseline,但是发现Unet量化结果奇高啊
是的,Unet就是最高的
— Reply to this email directly, view it on GitHubhttps://github.com/Beckschen/TransUNet/issues/110#issuecomment-1367757654, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AFQCHCWOP56DYGN46SHLF2TWP2B2LANCNFSM6AAAAAASS2XJ7I. You are receiving this because you authored the thread.Message ID: @.***>
本来想测试一下baseline,但是发现Unet量化结果奇高啊
怎么办,UNet精度太高,超不过baseline