6057CEM_10862829
This is the submission code for the module 6057CEM. The code develops an American Sign Language Recognition System using the MNIST ASL Dataset. CNN code created from scratch is presented in mnist1.py. CNN_DNN notebook provides an updated version of mnist1.py using multiple layers. The model is compared to a DNN approach and was tuned using grid search and tested using k-fold cross-validation. Evaluation metrics like confusion matrix and classfication report was produced. The comparitive analysis determined a multi-layer CNN approach to produce the most optimum accuracy results for ASL recognition on the MNIST dataset.