Open dudqls1994 opened 5 years ago
@dudqls1994 i wouldnt recommend changing anything in the code even if you are training it on the custom images, any way the filter size of a layer depends on the input image depth. if u still want to change the kernel size, change it in the size option in cfg file. example
[convolutional] batch_normalize=1 filters=32 size=3 #change this to change the kernel size stride=1 pad=1 activation=leaky
Hello! . Thank you for your code.
I have a question.
how to change convolution filter shape?? For example, In this network use 3x3 kernel. But I want to use 1X3 kernel. Thank you.