Open ersinsiginc opened 4 years ago
I have different objects on an image with the same class. But during training, images with "_gt" in the checkpoint folder show these objects as combined. What should I do to perceive them differently?
Maybe a problem of resizing + cropping. You can try to remove the copping in the train.py file.
I have different objects on an image with the same class. But during training, images with "_gt" in the checkpoint folder show these objects as combined. What should I do to perceive them differently?