Python OpenCV GPU Face Detection

Python wrapper for GPU CascadeClassifier, should work with OpenCV 2 and 3. Will fall back to CPU CascadeClassifier if CUDA isn't installed, but if the CPU version enough, just use stock OpenCV Python.

Installation

First install CUDA.

On Debian/Ubuntu, after downloading the .deb provided by NVIDIA:

sudo dpkg -i cuda-repo-ubuntu1404_7.5-18_amd64.deb # Filename will vary
sudo apt-get update && sudo apt-get install cuda
echo "export CUDA_HOME=/usr/local/cuda-7.5" >> ~/.bashrc
echo "export PATH=$PATH:$CUDA_HOME/bin" >> ~/.bashrc
echo "export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CUDA_HOME/lib64" >> ~/.bashrc

Build/install OpenCV 3 from source:

sudo apt-get update && sudo apt-get install -y \
    build-essential cmake git libgtk2.0-dev pkg-config \
    libavcodec-dev libavformat-dev libswscale-dev python-dev \
    python-numpy libtbb2 libtbb-dev libjpeg-dev libpng-dev
git clone https://github.com/Itseez/opencv
mkdir opencv/build && cd opencv/build
# Build OpenCV 2.4 instead of 3:
# mkdir opencv/build && cmake opencv && git checkout 2.4 && cd build
cmake .. && make -j$(nproc) && sudo make install

Install opencv-gpu-py:

git clone https://github.com/alexanderkoumis/opencv-gpu-py
pip install --user -e opencv-gpu-py

Test:

cd opencv-gpu-py
python test/test.py

Usage

# Full example in test/test.py
import cv2gpu

if cv2gpu.is_cuda_compatible():
    cv2gpu.init_gpu_detector('cascade_frontalface_gpu.xml')
else:
    cv2gpu.init_cpu_detector('cascade_frontalface_cpu.xml')

for (x, y, w, h) in cv2gpu.find_faces('image.jpg'):
    # Do something with face rectangle