You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As the method shows, it directly transform the bbox_delta (dx,dy,dw,dh) to pred_bbox (x1,y1,x2,y2) without the information of the ROIs. Obviously, it is impossible. Therefore, is there something I misunderstanding?
The text was updated successfully, but these errors were encountered:
Hi, I'm confusing the bbox_transform method in this file:
Detectron.pytorch/lib/utils/net.py
Line 15 in dcfab5f
As the method shows, it directly transform the bbox_delta (dx,dy,dw,dh) to pred_bbox (x1,y1,x2,y2) without the information of the ROIs. Obviously, it is impossible. Therefore, is there something I misunderstanding?
The text was updated successfully, but these errors were encountered: