Hi, I was checking out the code but I couldn’t understand how is loss calculated on heatmap.
Input size is [511,511] whereas the heatmap size is [128,128]. So, when calculating the detection loss, how can the gt corner locations be placed on smaller scale heatmap?
Hi, I was checking out the code but I couldn’t understand how is loss calculated on heatmap.
Input size is [511,511] whereas the heatmap size is [128,128]. So, when calculating the detection loss, how can the gt corner locations be placed on smaller scale heatmap?
Thank you.