This repository contains my implementations of deep neural networks on image and text datasets executed as part of ECE60146 Deep Learning coursework.
Focused on implementing several types of neural networks to perform multi-instance object classification, detection, and localization using the COCO dataset. Some of the tasks accomplished are developing custom datasets based on the goals, implementing Convolutional Neural Networks (CNN) and Vision Transformers (ViT) for image classification, implementing YOLO network with residual/skip blocks for object detection and localization for accurate bounding box prediction.
Performed sentiment analysis of Amazon reviews using Gated Recurrent Unit (GRU) networks and its variants for successful classification of reviews by analyzing their positive/negative sentiment.