Tianyu Yang

Please check my website: [tianyu-yang.com](https://tianyu-yang.com)

Just as @cotrane said, it still needs to add the flag `D_GLIBCXX_USE_CXX11_ABI=0` into the nvcc_complile.sh, right after the `$TF_INC`. `nvcc -std=c++11 -c -o deform_conv.cu.o deform_conv.cu.cc -I $TF_INC -D_GLIBCXX_USE_CXX11_ABI=0 -D GOOGLE_CUDA=1...

it is used to build the graph of filter and response which are defined in the following code.

- For the black image, I guess it should be some display problem. - I use python3.5 - I use python version of otb benchmark and convert the result to...

You need to train the network first to get the models for tracking inference.

Sorry for missing this link. You can download it from [here](http://www.vlfeat.org/matconvnet/models/imagenet-vgg-verydeep-16.mat). Have fixed it in readme.