I know, there are a lot of people asking errors due to a different size. I had this issue too, but I am training the model now.
I have to say, the author is great. We are using this tool because of him. However, this tool is not documented well enough to be easy to use.
Given the size error, you have to change every 480 and 360 to 224 and 224 (in my case). Also, you have to change a few numbers in def inference() in model.py. All deconv_layer have some numbers related to the size, such as 180 (360/2), 90 (360/2/2), and 45 (260/2/2/2). You have to replace them by yourself.
Hi.
I know, there are a lot of people asking errors due to a different size. I had this issue too, but I am training the model now.
I have to say, the author is great. We are using this tool because of him. However, this tool is not documented well enough to be easy to use.
Given the size error, you have to change every 480 and 360 to 224 and 224 (in my case). Also, you have to change a few numbers in
def inference()in model.py. Alldeconv_layerhave some numbers related to the size, such as 180 (360/2), 90 (360/2/2), and 45 (260/2/2/2). You have to replace them by yourself.