Pytorch implementation of yolov3?

[AI-Hunter]

Thanks for nice repo and your youtube explanation. I run your program, if I don't use pretrained model, where it initialized the weights ?

Thanks a lot

[Crazylov3]

By default, pytorch do it for you, automatically

[kjnam67]

Thank you for the details of the implementation. Very helpful. I ran the code and found the image shape was expected to be 3x416x416. But the image read from .jpg file was 416x416x3. Is there a reason to use the permuted shape in your code? Also, I tested the model using random numbers and passed. But, when I loaded real images, I get "RuntimeError: expected scalar type Byte but found Float." It's traced to be the following. Wondering if you had experience like this.

442 return F.conv2d(input, weight, bias, self.stride, --> 443 self.padding, self.dilation, self.groups)