Coordinate offset
#17
by
Wingogo182
- opened
Why are the bounding boxes inferred by the demo's image recognition always with scaling and offset? We have tried other images and the results were the same. Could you please tell us if the input image has undergone any preprocessing?