An Easy to install application of tensorflow object detection, using opencv. This is an implementation of the tensorflow object_detection research. Windows Support only
A web camera, or a USB web camera.
Preferably anaconda python 3.6 version: https://www.anaconda.com/download/#windows
This image is not optional, the library will not have the proper dependencies without it. To install a CPU version, one can skip these steps and simply run the setup.bat file if on windows.
Navigate to where environment.yml is located and run:
conda env create -f environment.yml #Should supply proper python image
conda activate tf15 #To activate the enviorment
After creating the environment, anaconda should automatically mount the new environment. From here follow the installation:
pip install opencv-python
pip install matplotlib
pip install image
pip install Cython
#Warning this will change your tensorflow version
pip install --ignore-installed --upgrade tensorflow==1.5
Or for GPU:
pip install --ignore-installed --upgrade tensorflow-gpu==1.5
It is important to note that the gpu version requires additional requirements see: https://www.tensorflow.org/install/install_windows
Remember to navigate into the correct file system first.
python main.py
Sometimes opencv gets confused, if multiple cameras are connected. If opencv is selecting the wrong camera, or is throwing errors, change cap = cv2.VideoCapture(0) to cap = cv2.VideoCapture(1) in main.py.