get empty predicted boxes when run 'tools/demo.py'

[zimenglan-sysu-512]

hi @daijifeng001 when i run the tools/demo.py, i print the pred_dict['boxes'], i get empty results as below:

Demo for data/demo/2008_000533.jpg
forward time 0.124833
pred_dict['boxes']:
[]

Demo for data/demo/2008_000910.jpg
forward time 0.194351
pred_dict['boxes']:
[]

Demo for data/demo/2008_001602.jpg
forward time 0.176894
pred_dict['boxes']:
[]

Demo for data/demo/2008_001717.jpg
forward time 0.171917
pred_dict['boxes']:
[]

Demo for data/demo/2008_008093.jpg
forward time 0.194658
pred_dict['boxes']:
[]

can you do me a favor? how to fix it?
thanks.

[Shehz]

Did you fix empty bounding box problem? Even when I ran the ./tools/demo.py, entire images are being segmented as background. Should I change any parameter? I followed every steps mentioned in readme to view output even then I'm not getting any result. Please let me know whether I'm doing anything wrong. @daijifeng001 . Also, I have downloaded model MNC 5 stage from the script given in reame file.

[wolf943134497]

_cls_score_0_split
I0103 14:58:24.866758 19156 net.cpp:865] Ignoring source layer rpn_loss_bbox
I0103 14:58:24.866789 19156 net.cpp:865] Ignoring source layer rpn_loss_cls
I0103 14:58:24.866992 19156 net.cpp:865] Ignoring source layer seg_cls_score_ext_seg_cls_score_ext_0_split
I0103 14:58:24.867004 19156 net.cpp:865] Ignoring source layer seg_cls_score_seg_cls_score_0_split

Demo for data/demo/2008_000533.jpg
forward time 0.166395

Demo for data/demo/2008_000910.jpg
forward time 0.224166

Demo for data/demo/2008_001602.jpg
forward time 0.215400

Demo for data/demo/2008_001717.jpg
forward time 0.203793

Demo for data/demo/2008_008093.jpg
forward time 0.202524
I have the same problem. but not print the pred_dict['boxes'].

[Guominyingxiongququ]

Did anyone solve this problem?

[mldm4]

Same problem here =(
I have debugged and checked that the problem is in the scores that the network outputs. They are close to 0 and do not overpass the threshold to be taken into account (variable vis_thres in get_vic_dict). So the problem is... why is the network behaving this way and not detecting the objects with good scores?? This is supposed to be solved in issue #46 but that solution is not working for me.