Devanagari-Script-Recognition

Hand written character recognition ML model to classify 48 characters of Devanagri script. Implemented 2D interpolation using NearestNDInterpolator to fill missing pixel information. Augmented the data by pixel shifting, increasing the size 4x, and improved the F1 score by 3% Achieved an F1 score of 0.983 by training an Ensemble of KNN, SVM and Random Forest classifiers