Closed Ahmed-M-Naguib closed 7 years ago
Hi, i'm trying to train the net on my own data and classes.
i'm getting assertion error in imdb.py line 120
the lines are:
boxes = self.roidb[i]['boxes'].copy() oldx1 = boxes[:, 0].copy() oldx2 = boxes[:, 2].copy() boxes[:, 0] = widths[i] - oldx2 - 1 boxes[:, 2] = widths[i] - oldx1 - 1 assert (boxes[:, 2] >= boxes[:, 0]).all()
i'm not sure. if the purpose is to flip image boxes, shouldn't it be like: boxes[:, 0] = widths[i] - oldx1 - 1 boxes[:, 2] = widths[i] - oldx2 - 1
since boxes[0] is x1, and boxes[2] is x2?
UPDATE: sorry, i just though about it. nevermind!
I have the same question, could you please tell me how to fix it ?
Hi, i'm trying to train the net on my own data and classes.
i'm getting assertion error in imdb.py line 120
the lines are:
boxes = self.roidb[i]['boxes'].copy() oldx1 = boxes[:, 0].copy() oldx2 = boxes[:, 2].copy() boxes[:, 0] = widths[i] - oldx2 - 1 boxes[:, 2] = widths[i] - oldx1 - 1 assert (boxes[:, 2] >= boxes[:, 0]).all()
i'm not sure. if the purpose is to flip image boxes, shouldn't it be like: boxes[:, 0] = widths[i] - oldx1 - 1 boxes[:, 2] = widths[i] - oldx2 - 1
since boxes[0] is x1, and boxes[2] is x2?
UPDATE: sorry, i just though about it. nevermind!