Processes a video or image from a camera source or file source, blurs a detected human face object in the image using a trained Haar cascade classifier while tracking that object to make sure all the frames are properly blurred. Made with real-time processing in mind.
Python 3 with NumPy and OpenCV 3.1+.
- Install Miniconda for your OS
- On Windows, use Anaconda Powershell Prompt as admin
- On Linux and Mac OS, install the .sh file and use the regular bash
- Install OpenCV
$ conda install -c conda-forge opencv
- Finally clone the repository
$ git clone https://github.com/VictorFSantolim/AutoFaceBlur.git
- Webcam input
$ python src/autofaceblur.py -c 0
- Image input and output
$ python src/autofaceblur.py -i assets/women.jpg -o assets/women_blur.jpg
- Video input and output, using the classifier we trained, and showing fps
$ python src/autofaceblur.py -s cascades/myCascade3.xml -v src_video_path -o blurred_video_path --show_fps
To close the window, press ESC.
-h or --help show the help message with all args and options and exit
-c CAMERA or --camera CAMERA select camera index CAMERA as the input source
-v VIDEO or --video VIDEO select path VIDEO as the video input source
-i IMAGE or --image IMAGE select path IMAGE as the image input source
-s CASCADE or --cascade_source CASCADE Defines CASCADE the trained Haar cascade classifier path. If the argument is absent, uses haarcascade_frontalface_default.xml.
-o OUTPUT_FILE or --output_file OUTPUT_FILE Enable output storing, saves the processed result to OUTPUT_FILE
-p or --hide_processing Runs processing in background, without displaying the frames being processed
--variable_fps Takes into account the processing time of the frame to track the object, only recommended for camera input.
--show_fps Shows fps in the a window while processing the video.
--save_fps Saves the video fps as it was captured to the video output_file.