Leverage ML to find images with similar faces on your local machine.
- Clone this repo
- Unzip the data/.bz2 files
- Ensure the following have been installed:
- dlib (https://github.com/davisking/dlib)
- tqdm (https://github.com/tqdm/tqdm)
- OpenCV (https://pypi.org/project/opencv-python/)
- NumPy (https://github.com/numpy/numpy)
- bz2file (https://pypi.org/project/bz2file/)
- Take a look at the example user code in main.py
- You can choose to use a HOG+SVM or a Max-Margin (MMOD) CNN face detector by setting the 'detector_type' argument to the Driver in main.py.
- The HOG+SVM detector provides accurate results and is computationally efficient. The CNN provides relatively better results but requires dlib with CUDA enabled to utilize your GPU
- Having filed in the required paths in main.py, run main.py from the root directory of this repo
- The HTML file containing the results is saved to the data folder by default as 'results.html'.