A generic face detection algorithm using haar-cascades and tweaking to improve the performance on raspberry pi 3.
A group project for our robotics lab, 6th Semester, Jan-April 2019
PROJECT STRUCTURE
RPI
- Figure out ip address
- โ The commands for transferring files from rpi to laptop
CODING
-
โ Face trainer
-
Haarcascade frontal face
-
LBPHFacerecognizer module
-
Image pipeline from folder os module
-
Map image from folder with image id
-
Grey-scale --> crop --> re-scale --> train --> save
DEBUGGING
- Opencv dependencies problems
- Rpi environment replicate on rpi
- Overcome build dlib problems
- Emulate environment on rpi
TWEAKING ( PERFORMANCE)
- Reduce resolution size to increase speed
- Improve dataset
References:
https://towardsdatascience.com/face-recognition-how-lbph-works-90ec258c3d6b
https://www.microsoft.com/en-us/research/project/msra-cfw-data-set-of-celebrity-faces-on-the-web/