Have implement neural network based mask detection algorithm with different colors of masks and that is working with 99% accuracy (around 2 fails in 850 tests) but the CNN based algorithm is too slow on boards like Raspberry Pi around 1-2 FPS at 720p. Shall I try PCA /ICA based techniques?
Yes, you can use Kernel PCA, ICA, LDA or multivariate regression. The fastest method is Bayesian framework of MLE or variational methods.
The Viola–Jones object detection framework is said to run in real-time and is not CNN/NN based.
In pca you need to decompose covariance matrix into eigenvalues and eigen vectors. And when your data dimensionality is high, its very time consuming.
Maybe hough transform based feature extraction algorithms help you.
https://www.researchgate.net › 2757...
Web results
(PDF) Face Recognition Using Hough Transform Based Feature Extraction.
The face detection can be implemented with OpenCV. Follow this link
https://towardsdatascience.com/face-detection-in-2-minutes-using-opencv-python-90f89d7c0f81
A bounding box is generated for each detected face. The method is fast and may work on Raspberry Pi.
https://mc.ai/covid-19%E2%80%8A-%E2%80%8Aface-mask-detector-using-deep-learning/
https://www.hindawi.com/journals/ddns/2020/9242917/
https://arxiv.org/pdf/2003.09093.pdf
"Detecting Masked Faces in the Wild with LLE-CNNs"
I think this article will help you.
Try using Farhan-Model as illustrated in the following article:
Article A new model for pattern recognition ✩
The source code of the model is available in both MATLAB and Python languages.
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing