naomichoy / face_blur

helping a friend to batch process photos

face_blur

This script modifies from https://github.com/ageitgey/face_recognition/blob/master/examples/ blur_faces_on_webcam.py and find_faces_in_picture_cnn.py to batch process photos in jpg/png format

HI CHING download the zip file, unzip and have fun.

DEPENDENCIES

Python 3.3+ is need to run the script. Google how to install python 3 on your mac/windows and follow the steps. (alternatively, install anaconda lmao)

You will also need to install Visual Studio before proceeding. (for compiling face_rcognition lib)

check if the following modules are installed. you need these modules to run the script

• numpy
• scipy
• opencv
• cmake (to install face_recognition)

pip install numpy scipy opencv-python cmake

you will also need to install this module I copied from ageitgey

pip install face_recognition

If you want to run the GPU version, install these

pip install numba cudatoolkit

Running the script

In your terminal, nevigate to the directory where you extracted the zip, run this command

python <filename>

idk maybe directly click and run the script might work too

updates

(17 Nov 2020) changed file names and added gpu version, not sure if it works :/

(verson 0.3 12 Nov 2020) metadata retaining required and added

(verson 0.2 10 Aug 2020) uploaded sample photos in corresponding "photos" and "processed" folders. Delete them before you use the code and put your photos in the photos folder.

(version 0.1)

-- create a folder called "photos" and a folder called "processed" within this unzipped folder. Put your photos in the photos folder.

-- rough script without flexible user directory input and options. some coding skills i've yet to learn. maybe i should include a file of my environment setup too

BRUH THIS IS SO SLOW / MY LAPTOP FROZE!

sorry this processing eats your computer resources, because pure CPU sucks at image processing. if your boss has a machine that has a NVIDIA GPU and dlib compiled with CUDA extensions (if he knows what this is he will understand this), the process can be accelerated

(edit -- i tried running the script with a laptop with a NVIDIA GPU and CUDA configured from previous projects and there's no speed up :( perhaps there's something i need to declare specifically to enable the GPU, otherwise you are gonna stick with the slow speed. better equip yourself with a good enough CPU)

https://www.geeksforgeeks.org/running-python-script-on-gpu/