ASL Image Classification using Computer Vision: Developed a Deep Learning model from scratch, based on Convolutional Networks using Python’s PyTorch libraries, to classify images of letters and numbers of the ASL
• Developed a Deep Learning model from scratch, based on Convolutional Networks using Python’s PyTorch libraries, to classify images of letters and numbers of the ASL, achieving a 96% accuracy score with a memory footprint of 1 GB after 40 epochs • Pre-processed images with data augmentation and dimension reduction with PyTorch methods like ImageDataGenerator, transforms, and data loader optimizing around 20% the memory usage • Built visualization for the images and the classification report of the model with sk-learn.
The image data is avaibalable in Kaggle to be downloaded in this link through Kaggle API: https://www.kaggle.com/datasets/ayuraj/asl-dataset