Armenian-Letters-Recognition
AlinaMuradyan/Armenian-Letters-Recognition
CNN
armenian-languagecharacter-recognitionmachine-learningconvolutional-neural-networkimage-classificationkerastensorflowjupyter-notebook
Summary
A CNN-based project for recognizing Armenian alphabet letters from grayscale images using the Mashtots Dataset v2. The repository contains a Jupyter Notebook implementing a convolutional neural network with two convolutional blocks, batch normalization, dropout, and callbacks for training optimization. It achieved approximately 91% validation accuracy.