Hi there! Thank you for your great work . I would like to know that output of after passing the image to img2feats function is 64x112x200.
In centernet the 64 implies C which is no. of object categories but in our case the number of object categories is 10.
Then why are the features channel 64?. The regression heads and heatmaps add upto 48
Also what does reg tells us about? If it is regression then have we add it to the dictionary list when the regression heads
are already then why add it
Hi there! Thank you for your great work . I would like to know that output of after passing the image to img2feats function is 64x112x200. In centernet the 64 implies C which is no. of object categories but in our case the number of object categories is 10.
Then why are the features channel 64?. The regression heads and heatmaps add upto 48
{'hm': 10, 'reg': 2, 'wh': 2, 'dep': 1, 'rot': 8, 'dim': 3, 'amodel_offset': 2, 'dep_sec': 1, 'rot_sec': 8, 'nuscenes_att': 8, 'velocity': 3}
Also what does reg tells us about? If it is regression then have we add it to the dictionary list when the regression heads are already then why add it