Closed Ivan-Zhao closed 6 years ago
This is due to an error in the model.py under the predict method. When the dimensions of the original image do not line up with the "img_size" the bottom right corner is not calculated. add this code to the predict method to fix the issue
if (x.shape[0]%self.img_size != 0 and x.shape[1]%self.img_size != 0):
tmp = self.sess.run(self.out,feed_dict={self.input:[x[-1*self.img_size:,-1*self.img_size:]]})[0]
tmp_image[-1*tmp.shape[0]:,-1*tmp.shape[1]:] = tmp
Another problem,how to set learning rate?
@PrathamSoni Thanks again, please view my comment on your pull request.
@Ivan-Zhao learning rate can be set in the code where AdamOptimizer is initialized. If you'd like, I can create a command line argument for it instead
I am also having this issue but I fixed it with the above code. This is an example of the issue.
Fix was merged.
The results of the network have black blocks in the lower right corner of the image