Segmentation fault (core dumped) when generating tf_records for R-Net

[waquey]

Hello, based on your tutorial step 10 "gen_RNet_tfrecords.py to generate tfrecords for RNet.(you should run this script four times to generate tfrecords of neg,pos,part and landmark respectively)", I've tried to generate for positive set at first.

"get_dataset"

item = '%s/pos_%s.txt' % (net,net) item = '%s/part_%s.txt' % (net,net) item = '%s/neg_%s.txt' % (net,net) item = '%s/landmark_%s_aug.txt' % (net,net)

However, it'll show the following messages after converting all images when generating tfrecords

Segmentation fault (core dumped)

Is there something I need to modify? Thanks

[fire717]

do u changed the saved tfrecord filename? if not, the four might be replaced into one file

[waquey]

Yes, I've tried to feed them as parameter for run() in gen_RNet_tfrecords.py

run(dir, net, output_directory, 'pos_landmark.tfrecord', ..) run(dir, net, output_directory, 'part_landmark.tfrecord', ..) run(dir, net, output_directory, 'neg_landmark.tfrecord', ..) run(dir, net, output_directory, 'landmark_landmark.tfrecord', ..)

However, the tfrecords files can be generated actually but just show "segmentation fault"

[LinShihChieh]

Hi, I have encountered such a same problem. Have you solve it? I found that the loop actually get this segmentation fault crash in tfrecord_writer.close() But I can't solve it.

[liuyunwww]

#14

[liuyunwww]

https://github.com/AITTSMD/MTCNN-Tensorflow/issues/3