A MATLAB toolbox for Photoplethysmography Imaging
A lightweight, heuristic-based algorithm for segmenting characters in Iranian license plates using OpenCV. Features robust handling of shadows, noise, and connected characters without Deep Learning, achieving 98.68% accuracy.
Engineering-oriented OCR pipeline focused on robustness and stability
The CVLCore project is an opportunity for continuous analysis of a video stream with the functionality of calibrating parameters to generate a vibro-image, including calculating statistics of changes in vibrating pixels.
Goal of the project is to recognize license plates on images without using deep learning methods. Project was prepared for course "Vision systems". Final accuracy on test set was around 80-85%.
To associate your repository with the no-deep-learning topic, visit your repo's landing page and select "manage topics."