-
Complete the pre-requisites
-
Clone Fast-RCNN and its Caffe fork ( It's better to do it in your home dir. CMake files will be looking for it on user's home)
% git clone --recursive https://github.com/rbgirshick/fast-rcnn.git
-
Go into
fast-rcnn/caffe-fast-rcnn
and run% cp Makefile.config.example Makefile.config
-
Open
Makefile.config
with your favorite editor.
- If you have cuDNN installed uncomment the line
# USE_CUDNN := 1
- Make sure to uncomment
WITH_PYTHON_LAYER := 1
- Compile Caffe
make && make distribute
- Download pretrained models executing while in
fast-rcnn
folder.
% ./data/scripts/fetch_fast_rcnn_models.sh
or http://www.cs.berkeley.edu/~rbg/fast-rcnn-data/voc12_submission.tgz
and put it in fast-rcnn/data/fast_rcnn_models
.
To take advantage of cuDNN, at least CUDA 7.0 and a GPU with 3.5 compute capability is required.
If you didn't install Fast-RCNN in your home, modify the librcnn and rcnn_node CMakeFiles and point them to your caffe directories.
Once compiled, if you are running from the terminal
% roslaunch cv_tracker rcnn.launch
Remember to modify the launch file located inside computing/perception/detection/packages/cv_tracker/launch/rcnn.launch
and point the network and pretrained models to your path