Mexican Sign Language recognition (OpenCV + Keras)
Real-time recognition of the LSM hand-sign alphabet from a webcam
A computer-vision and deep-learning project (February to May 2019): a system that watches a webcam feed and classifies the hand sign being shown into a letter of the Lengua de Señas Mexicana (LSM) alphabet.
I researched the LSM alphabet, captured and labeled a custom dataset, then modeled and trained a CNN in Keras — using OpenCV for the live capture and preprocessing pipeline. The end result classified the alphabet in real time on a laptop webcam.