Py-faster-rcnn: Can't make test! fatal error: caffe/vision_layers.hpp: No such file or directory

Created on 26 Mar 2016 · 5Comments · Source: rbgirshick/py-faster-rcnn

Hi,

I am trying to run 'make test' in caffe-fast-rcnn, but receive the following error during the make:

src/caffe/test/test_smooth_L1_loss_layer.cpp:11:35: fatal error: caffe/vision_layers.hpp: No such file or directory

I have read on 2 threads that I this occurs with newer versions of caffe, and the recommendations are to revert to an older version, however I'm not sure this is possible with the Faster-RCNN version of caffe.

Would be very grateful if anyone could advise me on how to proceed!

Thanks in advance.

Source

boyentenbi

Most helpful comment

Hi,
I had the same issue, then simply removed src/caffe/test/test_smooth_L1_loss_layer.cpp file from running by renaming it to test_smooth_L1_loss_layer.cpp.orig.
This issue should be the only problem in make testand in make runtest. If both pass, you can expect the demo to run smoothly on your GPU.

davidkh1 on 27 Mar 2016

👍17 🎉2

All 5 comments

davidkh1 on 27 Mar 2016

👍17 🎉2

Hi @davidkh1 I did what you said, but still failed in ./tools/demo.py

...

F0707 17:44:18.173039  5377 layer_factory.hpp:81] Check failed: registry.count(type) == 1 (0 vs. 1) Unknown layer type: Python (known types: AbsVal, Accuracy, ArgMax, BNLL, BatchNorm, BatchReindex, Bias, Concat, ContrastiveLoss, Convolution, Data, Deconvolution, Dropout, DummyData, ELU, Eltwise, Embed, EuclideanLoss, Exp, Filter, Flatten, HDF5Data, HDF5Output, HingeLoss, Im2col, ImageData, InfogainLoss, InnerProduct, LRN, Log, MVN, MemoryData, MultinomialLogisticLoss, PReLU, Pooling, Power, ROIPooling, ReLU, Reduction, Reshape, SPP, Scale, Sigmoid, SigmoidCrossEntropyLoss, Silence, Slice, SmoothL1Loss, Softmax, SoftmaxWithLoss, Split, TanH, Threshold, Tile, WindowData)
*** Check failure stack trace: ***
Aborted (core dumped)

any idea to fix this?

xuyuan on 7 Jul 2016

Hi,
Unknown layer type: Python
This error indicates that WITH_PYTHON_LAYER := 1 flag in Makefile.config was not set. Uncomment it to support Python layers and recompile Caffe.

davidkh1 on 8 Jul 2016

👍3

Hi @davidkh1 demo works great, thank you!

xuyuan on 8 Jul 2016

I confirmed the error after I rebase the branch on to the latest caffe version and fixed cuda conflictions with caffe main branch. The test file should be removed. B.T.W, codes of py-faster-rcnn was written in python2. Do not compile the codes using python 3. Environment, Ubuntu & CentOS system. The issue can be closed now.