I’m using pytorch 1.3.0 with torchvision 0.4.1. I try to do RoIAlign with torchvision.ops.roi_align and I’m not sure what’s the meaning of the ‘spatial_ratio’ parameter. As described in the documentation, this parameter should map the input coordinates to the box coordinates. Normally, input coordinates are in the feature space and box coordinates are in the image space. So does this mean that this parameter should be greater than one?
Hi, did you found any explanation about it ?