caffe
caffe copied to clipboard
Missing true_pos for label
I changed proto, but I got this message. why?
W0911 00:43:53.443533 20747 solver.cpp:512] Missing true_pos for label: 1 W0911 00:43:53.443758 20747 solver.cpp:512] Missing true_pos for label: 2 W0911 00:43:53.443768 20747 solver.cpp:512] Missing true_pos for label: 3 W0911 00:43:53.443771 20747 solver.cpp:512] Missing true_pos for label: 4 W0911 00:43:53.443778 20747 solver.cpp:512] Missing true_pos for label: 5 W0911 00:43:53.443781 20747 solver.cpp:512] Missing true_pos for label: 6 W0911 00:43:53.514109 20747 solver.cpp:512] Missing true_pos for label: 8 W0911 00:43:53.515017 20747 solver.cpp:512] Missing true_pos for label: 10 W0911 00:43:53.515419 20747 solver.cpp:512] Missing true_pos for label: 12 W0911 00:43:53.515429 20747 solver.cpp:512] Missing true_pos for label: 13 W0911 00:43:53.515434 20747 solver.cpp:512] Missing true_pos for label: 14 W0911 00:43:53.529917 20747 solver.cpp:512] Missing true_pos for label: 16 W0911 00:43:53.529930 20747 solver.cpp:512] Missing true_pos for label: 17 W0911 00:43:53.529935 20747 solver.cpp:512] Missing true_pos for label: 18 W0911 00:43:53.529940 20747 solver.cpp:512] Missing true_pos for label: 19 W0911 00:43:53.529944 20747 solver.cpp:512] Missing true_pos for label: 20
What exactly have you changed? Without more detailed information, it is hard for me to know what the problem is.
Seems like the name of object classes are different in training data/lmdb and the labelmap file ?
@weiliu89
I use my own dataset to train the SSD, and this happens only in testing and scoring the mAP, where could the problem be?
and I have no problem when running https://github.com/weiliu89/caffe/blob/ssd/examples/ssd_detect.ipynb
@jtr2012 if you have solve it ? i also have same problem