Description: This is a beginner computer vision (CV) project to introduce participants to the basics of modern computer vision ideas within deep learning.
Participants will begin by implementing one or two classical machine learning (ML) models such as logistic regression, k-nearest neighbors (KNN), support vector machines (SVM), etc. for digit classification with the MNIST dataset. This foundational step will provide an understanding of the basics of feature extraction, classification, and the limitations of classical models for processing structured image data.
Building on the knowledge gained from implementing non-deep learning models, participants will transition to implementing a convolutional neural network (CNN) using PyTorch. They will learn to design and train a CNN for image recognition tasks. Emphasis will be placed on gaining real life practice by writing your own models, datasets and dataloaders, training/testing/validation loops, and understanding the intuition behind the CNN architecture and its effectiveness in processing structured image data.
Bi-weekly meetings will take place to ensure progress and to provide support.
Time commitment per week: 7 hours minimum + 30 minutes for bi-weekly meetings
Estimated length: 2 months
Group max size: 3
Current members: