1 | initial version |
Forget my previous comment, which I deleted because it was plain wrong
Taking a look at the negative image grabber you can see that the scaling step of the negative samples is defined by the model dimensions and the image dimensions provided. The exact line of defining this scale step is found here.
2 | No.2 Revision |
Forget my previous comment, which I deleted because it was plain wrong
Taking a look at the negative image grabber you can see that the scaling step of the negative samples is defined by the model dimensions and the image dimensions provided. The exact line of defining this scale step is found here.
The scaleFactor itself is hardcoded right here and is used to define the new scale for each step.