tool/darknet2pytorch.py 不执行任何代码

[AmazingYi]

大佬你好，在你的readme 文件中说明用darknet2pytoech 转化原有权重模型。但是这个py文件只有几个类，并不会执行任何代码，而且还有tool import 错误。请教一下大佬原本的思路是什么呢？

[RuilongMachineLearning]

需要自己存一下：新建一个 .py file: from tool import darknet2pytorch import torch

model = darknet2pytorch.Darknet('./cfg/yolov4.cfg', inference=True) model.load_weights('./weights/yolov4.weights')

torch.save(model.state_dict(), './weights/yolov4-pytorch.pth')

[dhakalkamal]

Actually this shows the error as Convalution haven't activation layer.

by the way to How to convert yolov4.conv.137.weights into the .pth format?

[myrzx]

需要自己存一下：新建一个 .py file: from tool import darknet2pytorch import torch

model = darknet2pytorch.Darknet('./cfg/yolov4.cfg', inference=True) model.load_weights('./weights/yolov4.weights')

torch.save(model.state_dict(), './weights/yolov4-pytorch.pth')

你好，我使用这段代码将yolov4.weights转化为pth文件，但运行时出现了如下错误 in load_state_dict RuntimeError: Error(s) in loading state_dict for Yolov4: Missing key(s) in state_dict : "dow1.conv1.conv.0.weight", "down1.con1.conv.1.weight", "down1.conv1.conv.1.bias" ....... Unexpected Key(s) in state_dict: "models.0.conv1.weight", "models.0.bnl.weight", "models.0.bnl.bias"

[wj1017090777]

@myrzx 你好，我也出现了该问题，请问解决了吗

[dioptrique]

需要自己存一下：新建一个 .py file: from tool import darknet2pytorch import torch model = darknet2pytorch.Darknet('./cfg/yolov4.cfg', inference=True) model.load_weights('./weights/yolov4.weights') torch.save(model.state_dict(), './weights/yolov4-pytorch.pth')

你好，我使用这段代码将yolov4.weights转化为pth文件，但运行时出现了如下错误 in load_state_dict RuntimeError: Error(s) in loading state_dict for Yolov4: Missing key(s) in state_dict : "dow1.conv1.conv.0.weight", "down1.con1.conv.1.weight", "down1.conv1.conv.1.bias" ....... Unexpected Key(s) in state_dict: "models.0.conv1.weight", "models.0.bnl.weight", "models.0.bnl.bias"

I am facing exactly the same issue when converting darknet weights on a yolov4 model with 2 classes to pytorch weights. All the key names seem to be different @Tianxiaomo

[RuochenQi]

@shuuny-matrix Convalution haven't activation layer merely a warning, meaning no layer named as activation this issue has been discussion before, search in this git should find it