In the file matting_unpool.py, to predict the values for validation pred_final variable is run whereas in test.py for testing pred_mattes is run. Isn't this something weird. I'm getting different results for a single example during validation, test modes. Why pred_mattes is added to the collection not pred_final?
In the file matting_unpool.py, to predict the values for validation pred_final variable is run whereas in test.py for testing pred_mattes is run. Isn't this something weird. I'm getting different results for a single example during validation, test modes. Why pred_mattes is added to the collection not pred_final?