error showed up when tried to run run_pascal_strong.sh

[daofeng2007]

* Aborted at 1470496762 (unix time) try "date -d @1470496762" if you are using GNU date * PC: @ 0x7f1e2d62bf20 (unknown) * SIGSEGV (@0x0) received by PID 2899 (TID 0x7f1e35828800) from PID 0; stack trace: * @ 0x7f1e2c887cb0 (unknown) @ 0x7f1e2d62bf20 (unknown) @ 0x51f94c std::operator+<>() @ 0x52170d caffe::ImageSegDataLayer<>::DataLayerSetUp() @ 0x5341aa caffe::BaseDataLayer<>::LayerSetUp() @ 0x5349a1 caffe::ImageDimPrefetchingDataLayer<>::LayerSetUp() @ 0x499669 caffe::Net<>::Init() @ 0x49b07e caffe::Net<>::Net() @ 0x4e5229 caffe::Solver<>::InitTrainNet() @ 0x4e6586 caffe::Solver<>::Init() @ 0x4e66e6 caffe::Solver<>::Solver() @ 0x41e460 caffe::GetSolver<>() @ 0x419fa4 train() @ 0x412ba1 main @ 0x7f1e2c872f45 (unknown) @ 0x418ae7 (unknown) @ 0x0 (unknown) Segmentation fault (core dumped) I used GTX 960 2GB to run this model, generating the error above, any suggestion?

[daofeng2007]

The error above disappeared when I used the latest caffe-master instead of using the one provided by deeplab. However, new errors showed up when I tried to run run_pascal_strong.sh

I0806 13:06:51.480073 3384 solver.cpp:81] Creating training net from train_net file: exper/voc12/config/DeepLab-LargeFOV/train_trainval_aug.prototxt [libprotobuf ERROR google/protobuf/text_format.cc:298] Error parsing text-format caffe.NetParameter: 24:3: Unknown enumeration value of "IMAGE_SEG_DATA" for field "type". F0806 13:06:51.480228 3384 upgrade_proto.cpp:79] Check failed: ReadProtoFromTextFile(param_file, param) Failed to parse NetParameter file: exper/voc12/config/DeepLab-LargeFOV/train_trainval_aug.prototxt * Check failure stack trace: * @ 0x7f31a5a8ddaa (unknown) @ 0x7f31a5a8dce4 (unknown) @ 0x7f31a5a8d6e6 (unknown) @ 0x7f31a5a90687 (unknown) @ 0x7f31a60d68ee caffe::ReadNetParamsFromTextFileOrDie() @ 0x7f31a611dc47 caffe::Solver<>::InitTrainNet() @ 0x7f31a611ec9c caffe::Solver<>::Init() @ 0x7f31a611efca caffe::Solver<>::Solver() @ 0x7f31a6117693 caffe::Creator_SGDSolver<>() @ 0x40f56d caffe::SolverRegistry<>::CreateSolver() @ 0x408664 train() @ 0x405f3c main @ 0x7f31a4042f45 (unknown) @ 0x406843 (unknown) @ (nil) (unknown) Aborted (core dumped)

[martinkersner]

Hi @daofeng2007,

You should use the Caffe which is located in this repository (code directory). In order to get it you have to clone this repository with --recursive option.

Cheers,

Martin

[maohule]

@daofeng2007 Have you solve your problem.I encounter same problem.

[dxiaosa]

@daofeng2007 @martinkersner I have the same problem, so what's wrong?

[libprotobuf ERROR google/protobuf/text_format.cc:298] Error parsing text-format caffe.NetParameter: 24:3: Unknown enumeration value of "IMAGE_SEG_DATA" for field "type". F0108 22:52:08.534060 8636 upgrade_proto.cpp:88] Check failed: ReadProtoFromTextFile(param_file, param) Failed to parse NetParameter file: exper/voc12/config/DeepLab-LargeFOV/train_train_aug.prototxt

[mkiernan]

Same problem on Ubuntu 16.04 despite following the instructions to the letter.

Anyone have a workaround ?

state { phase: TRAIN } I0319 07:26:46.559741 66719 layer_factory.hpp:78] Creating layer data I0319 07:26:46.559768 66719 net.cpp:67] Creating Layer data I0319 07:26:46.559788 66719 net.cpp:356] data -> data I0319 07:26:46.559813 66719 net.cpp:356] data -> label I0319 07:26:46.559828 66719 net.cpp:356] data -> (automatic) I0319 07:26:46.559839 66719 net.cpp:96] Setting up data I0319 07:26:46.559846 66719 image_seg_data_layer.cpp:45] Opening file exper/voc12/list_subset/train_aug.txt I0319 07:26:46.559888 66719 image_seg_data_layer.cpp:62] Shuffling data I0319 07:26:46.560199 66719 image_seg_data_layer.cpp:67] A total of 0 images. Aborted at 1489908406 (unix time) try "date -d @1489908406" if you are using GNU date PC: @ 0x5347a3 std::operator+<>() SIGSEGV (@0x8) received by PID 66719 (TID 0x7f70a232bac0) from PID 8; stack trace: @ 0x7f7098064390 (unknown) @ 0x5347a3 std::operator+<>() @ 0x54d8e2 caffe::ImageSegDataLayer<>::DataLayerSetUp() @ 0x51d9d5 caffe::BaseDataLayer<>::LayerSetUp() @ 0x51dd01 caffe::ImageDimPrefetchingDataLayer<>::LayerSetUp() @ 0x5adfde caffe::Net<>::Init() @ 0x5afb8a caffe::Net<>::Net() @ 0x589cb5 caffe::Solver<>::InitTrainNet() @ 0x58af82 caffe::Solver<>::Init() @ 0x58b0d6 caffe::Solver<>::Solver() @ 0x4214f0 caffe::GetSolver<>() @ 0x41c1fa train() @ 0x415c28 main @ 0x7f7097caa830 __libc_start_main @ 0x41ac89 _start @ 0x0 (unknown) Segmentation fault (core dumped)

[Bob-Yeah]

Yes. I have the same problem with @mkiernan . There two practice here VOC original and VOC aug. You can comment the TRAIN_SET_SUFFIX=_subset and uncomment the TRAIN_SET_SUFFIX= in the run_pascal_strong.sh to run the training of VOC original dataset.

But then I encounter with the out of memory error. How much memory should I prepare? I'm working on my laptop with 8G mem and GTX 970M with 3GB mem ............Sigh~

F1123 19:37:44.117923 17058 syncedmem.cpp:51] Check failed: error == cudaSuccess (2 vs. 0) out of memory