zzingae

Results 3 comments of zzingae

My test results of horse2zebra is somewhat different from this.

I found that each training with the same code can lead to very different result due to the difference of initial weight values. I was able to obtain very similar...

There appear to be 1 leaked semaphore objects to clean up at shutdown

It seems related to device type (Mac mps type). When I move mps type tensor to cpu(), the problem no longer appears.

Great job..! I have a question for finding small objects

Thank you for your interest. Faster RCNN would be better to detect small objects than other detection networks because individual candidate is passed through fully connected layers. However, it would...