search

rbgirshick / py-faster-rcnn

Faster R-CNN (Python implementation) -- see https://github.com/ShaoqingRen/faster_rcnn for the official MATLAB version
Other
8.11k stars 4.11k forks source link

Train the VGG16 model by using my own data #411

Open learntolearn11 opened 7 years ago

learntolearn11 commented 7 years ago

I1118 09:22:39.508401 44496 net.cpp:228] conv2_2 does not need backward computation. I1118 09:22:39.508407 44496 net.cpp:228] relu2_1 does not need backward computation. I1118 09:22:39.508411 44496 net.cpp:228] conv2_1 does not need backward computation. I1118 09:22:39.508416 44496 net.cpp:228] pool1 does not need backward computation. I1118 09:22:39.508421 44496 net.cpp:228] relu1_2 does not need backward computation. I1118 09:22:39.508425 44496 net.cpp:228] conv1_2 does not need backward computation. I1118 09:22:39.508431 44496 net.cpp:228] relu1_1 does not need backward computation. I1118 09:22:39.508435 44496 net.cpp:228] conv1_1 does not need backward computation. I1118 09:22:39.508442 44496 net.cpp:228] gt_boxes_input-data_n_2_split does not need backward computation. I1118 09:22:39.508448 44496 net.cpp:228] im_info_input-data_n_1_split does not need backward computation. I1118 09:22:39.508455 44496 net.cpp:228] data_input-data_n_0_split does not need backward computation. I1118 09:22:39.508460 44496 net.cpp:228] input-data_n does not need backward computation. I1118 09:22:39.508465 44496 net.cpp:270] This network produces output loss_bbox I1118 09:22:39.508471 44496 net.cpp:270] This network produces output loss_cls I1118 09:22:39.508476 44496 net.cpp:270] This network produces output rpn_cls_loss I1118 09:22:39.508481 44496 net.cpp:270] This network produces output rpn_loss_bbox I1118 09:22:39.508545 44496 net.cpp:283] Network initialization done. I1118 09:22:39.508874 44496 solver.cpp:60] Solver scaffolding done. Loading pretrained model weights from data/imagenet_models/VGG16.v2.caffemodel [libprotobuf WARNING google/protobuf/io/coded_stream.cc:505] Reading dangerously large protocol message. If the message turns out to be larger than 2147483647 bytes, parsing will be halted for security reasons. To increase the limit (or to disable these warnings), see CodedInputStream::SetTotalBytesLimit() in google/protobuf/io/coded_stream.h. [libprotobuf WARNING google/protobuf/io/coded_stream.cc:78] The total number of bytes read was 548317115 I1118 09:22:40.616070 44496 net.cpp:816] Ignoring source layer data I1118 09:22:40.726676 44496 net.cpp:816] Ignoring source layer cls_score I1118 09:22:40.726713 44496 net.cpp:816] Ignoring source layer bbox_pred I1118 09:22:40.728744 44496 net.cpp:816] Ignoring source layer silence_rpn_cls_score I1118 09:22:40.728752 44496 net.cpp:816] Ignoring source layer silence_rpn_bbox_pred Solving...

Hi, I am trainning the VGG16model by using my own data, but the process is stop at the Solving step and do not going on next step, does angone know how to fix this?

marifnst commented 7 years ago

me too. any solution ?

i get the same condition in low resource gpu, but run normally in my server.

learntolearn11 commented 7 years ago

hi I run it on my server which contains k80, while it still not work, it work right for you?

mantoone commented 7 years ago

What command/script did you run for training?

learntolearn11 commented 7 years ago

./experiments/scripts/faster_rcnn_end2end.sh 0 VGG16 pascal_voc