Train Custom Dataset
xv994 opened this issue · 8 comments
Hey, I am coming too. infer_mdt.py works now, but when I trained my dataset, I had a new problem as below.
Do you guys know why it has a segmentation error? Please help me!
When I ran the MDTv1, I met the same problem.
My GPU is 3090 24GB. batch_size = 1
It seems like a environment issue. Can you try to reinstall your pytorch environment and also the cuda and cudnn?
Can I add you to WeChat communication?
Can I add you to WeChat communication?
give me your email address or wechat account
Can I add you to WeChat communication?
give me your email address or wechat account
Can I add you to WeChat communication?
give me your email address or wechat account
已发送
Hey, I want to train on my custom dataset. But its unconditional data set, just the images, no class id no annotations. Can I still train? How can I do that?
@gasvn @xv994 @liwenyang-911 @mavenlin @P2333 Can you please help me out here if you know?
Hey, I want to train on my custom dataset. But its unconditional data set, just the images, no class id no annotations. Can I still train? How can I do that?
@gasvn @xv994 @liwenyang-911 @mavenlin @P2333 Can you please help me out here if you know?
I'm sorry for replying late.
This model is a supervised deep learning. Maybe using conditional dataset is not a right choice.