Unknow segmentation fault

[lucasjinreal]

Hi, train the encoder :+1:

caffe-enet/build/tools/caffe train -solver prototxts/enet_solver_encoder.prototxt

an I just encountered a very wired error:

I0112 21:06:24.985251 17605 layer_factory.hpp:77] Creating layer drop1_0_3
*** Aborted at 1515762385 (unix time) try "date -d @1515762385" if you are using GNU date ***
PC: @     0x7f0c4dbb5d53 std::_Hashtable<>::clear()
*** SIGSEGV (@0x9) received by PID 17605 (TID 0x7f0d086ed740) from PID 9; stack trace: ***
    @     0x7f0d055644b0 (unknown)
    @     0x7f0c4dbb5d53 std::_Hashtable<>::clear()
    @     0x7f0c4dbae4d6 google::protobuf::DescriptorPool::FindFileByName()
    @     0x7f0c4db64708 google::protobuf::python::cdescriptor_pool::AddSerializedFile()
    @     0x7f0d05e0d9e0 PyEval_EvalFrameEx
    @     0x7f0d05f4304c PyEval_EvalCodeEx
    @     0x7f0d05e9945d (unknown)
    @     0x7f0d05e6c263 PyObject_Call
    @     0x7f0d05e8cb65 (unknown)
    @     0x7f0d05e23163 (unknown)
    @     0x7f0d05e6c263 PyObject_Call
    @     0x7f0d05e0a34c PyEval_EvalFrameEx
    @     0x7f0d05f4304c PyEval_EvalCodeEx
    @     0x7f0d05e04d99 PyEval_EvalCode
    @     0x7f0d05ea6234 PyImport_ExecCodeModuleEx
    @     0x7f0d05ea6c0f (unknown)
    @     0x7f0d05ea8380 (unknown)
    @     0x7f0d05ea8648 (unknown)
    @     0x7f0d05ea975b PyImport_ImportModuleLevel
    @     0x7f0d05e138a8 (unknown)
    @     0x7f0d05e6c263 PyObject_Call
    @     0x7f0d05f42477 PyEval_CallObjectWithKeywords
    @     0x7f0d05e087d6 PyEval_EvalFrameEx
    @     0x7f0d05f4304c PyEval_EvalCodeEx
    @     0x7f0d05e04d99 PyEval_EvalCode
    @     0x7f0d05ea6234 PyImport_ExecCodeModuleEx
    @     0x7f0d05ea6c0f (unknown)
    @     0x7f0d05ea8380 (unknown)
    @     0x7f0d05ea8648 (unknown)
    @     0x7f0d05ea975b PyImport_ImportModuleLevel
    @     0x7f0d05e138a8 (unknown)
    @     0x7f0d05e6c263 PyObject_Call
[1]    17605 segmentation fault (core dumped)  caffe-enet/build/tools/caffe train -solver 

Really don't know why, even not any error throw out. I build caffe-enet successfully, and using protobuf3.5, Does anybody knows why???

[ewen1024]

first of all, layer 1_0_3 is the spatial dropout layer implemented by cpu. plz make sure you are doing right with the python script

[mingx9527]

The Error is caused by data. you can check your cityscapes data, no have groundtruth: xxx_xxx_xxx_gtFine_labelTrainIds.png you must: git clone https://github.com/mcordts/cityscapesScripts.git and then run the /preparation/createTrainIdLabelImags.py to generate xxx_xxx_xxx_gtFine_labelTrainIds.png As shown in the Tutorial