Build issue with OpenCV

[Ranjan Anantharaman]

I built MXNet with CUDA and OpenCV, and whenever I read a .rec file, I get the following issue:

julia> using MXNet
libdc1394 error: Failed to initialize libdc1394

julia> train_provider = mx.ImageRecordProvider(path_imgrec = filenames[:train], data_shape = (2048, 1536,3), batch_size = 10)
ERROR: UndefVarError: filenames not defined

julia> filenames = Dict(:train => "data/train.rec", :test => "data/test.rec")
Dict{Symbol,String} with 2 entries:
  :test  => "data/test.rec"
  :train => "data/train.rec"

julia> train_provider = mx.ImageRecordProvider(path_imgrec = filenames[:train], data_shape = (2048, 1536,3), batch_size = 10)
[07:32:23] src/io/iter_image_recordio.cc:209: ImageRecordIOParser: data/train.rec, use 3 threads for decoding..
MXNet.mx.MXDataProvider(MXNet.mx.MX_DataIterHandle(Ptr{Void} @0x00000000031b7be0),Tuple{Symbol,Tuple}[(:data,(2048,1536,3,10))],Tuple{Symbol,Tuple}[(:softmax_label,(10,))],10,true,true)

julia> terminate called recursively
terminate called recursively
terminate called after throwing an instance of 'std::bad_alloc'
  what():  std::bad_alloc

signal (6): Aborted
while loading no file, in expression starting on line 0
while loading no filewhile loading no file, in expression starting on line 0
while loading no filewhile loading no file, in expression starting on line 0
Allocations: 3486735 (Pool: 3485227; Big: 1508); GC: 4
Allocations: 3486735 (Pool: 3485227; Big: 1508); GC: 4
Aborted (core dumped)

Here's my versioninfo():

Julia Version 0.5.0-rc2+0
Commit 0350e57 (2016-08-12 11:25 UTC)
Platform Info:
  System: Linux (x86_64-unknown-linux-gnu)
  CPU: Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Sandybridge)
  LAPACK: libopenblas64_
  LIBM: libopenlibm
  LLVM: libLLVM-3.7.1 (ORCJIT, ivybridge)

cc: @vchuravy

[Chiyuan Zhang]

Does this help?

I vaguely remember I had some similar issue with lib1394. But I do not remember how I resolve it. Maybe I just compiled OpenCV by disabling everything (CUDA, video, etc.) that I thought would not be needed.

[Valentin Churavy]

@ranjanan I think we solved this, do you remember how?