Suggestions to Improve the Speed

ageitgey / face_recognition

The world's simplest facial recognition api for Python and the command line

MIT License

52.85k stars 13.43k forks source link

Suggestions to Improve the Speed #1250

Open psychosys opened 3 years ago

psychosys commented 3 years ago

How to implement multi processing to face detection and recognition of the image files ?

CPU option is kind of useless it still uses one single processing core! CUDA enabled dLib works nice but I don't know how to optimize the CPU part or split the task between multi cores

Thanks

bohdansok commented 3 years ago

Hi, You may want to use concurrent.futures to implement multithread approach to the heaviest math... Like I've done it here: https://github.com/bohdansok/Face_Recognition In fact. app works up to 3 times faster even on CPU (in my case) but - iy's not only after Mt enabled but a;sp due to further "traditional" code optimsation too..

zoldaten commented 2 months ago

use different detector - https://ai.google.dev/edge/mediapipe/solutions/vision/face_detector#models