search

VisionLearningGroup / MMVD

Multimodal Video Description
3 stars 3 forks source link

Question on C3D Feature extraction #1

Open medhini opened 7 years ago

medhini commented 7 years ago

I read your paper, "Multi modal Video Description" and learned that you have used the C3D network (pre-trained on Sports1M) for extraction action recognition features. Did you use the facebook/C3D repo to do this? If so, could you help me with the following issue?

I get this error while running "make all":

In file included from ./include/caffe/vision_layers.hpp:13:0, from src/caffe/layers/tanh_layer.cpp:9: ./include/caffe/neuron_layers.hpp:13:18: fatal error: hdf5.h: No such file or directory compilation terminated.

Makefile:325: recipe for target 'build/src/caffe/layers/tanh_layer.o' failed make: *** [build/src/caffe/layers/tanh_layer.o] Error 1

How do I go about fixing this? It would be of great help if you could release the code you used for extracting C3D features.

Peratham commented 7 years ago

I am just passing by but I think I could provide you a pointer.

Your error seems to be from caffe so I think you should look into some threads in https://github.com/BVLC/caffe about how to compile caffe.

czgdp1807 commented 5 years ago

I also read the paper and I also faced this issue while using caffe for C3D in another project. So, I am too late, and this issue would have been solved till now, but still if someone faces the same issue again, can try the discussion on this stack overflow page. The idea used was quite amazing in the paper. :-)

Villjoie commented 4 years ago

was this problem solved, please? if so, then could you help me?

I already compile Caffe which works very well, but when I compile Makefile in the directory /C3D-master/C3D-v1.0,

I had the same mistake.