No description available

eveloped a 3D robotics simulation in Webots to generate synthetic image datasets and trained a CNN using TensorFlow for real-time object classification and perception.

A deep learning project focused on reconstructing 3D faces from 2D images, with real-time feedback for dynamic visualization. Using frameworks like PyTorch, TensorFlow, and CNNs, this model processes each video frame to produce real-time 3D facial representations.

This project implements a 3D Umpire Review System using a Convolutional Neural Network (CNN) built with TensorFlow and Keras. The goal of the system is to classify decisions as "out" or "not out" based on cricket imagery. The model utilizes a binary classification approach and processes images of various situations during a cricket match.

Built a neural network from scratch in without using any standard libraries like tensorflow and keras for image classification Implemented a 2D convolution framework(both forward and backward propagation) with activation(ReLU) and Softmax Used a 3D CNN Sequential model(built in TensorFlow and Keras)for validating the 2D model predictions on CIFAR-10 dataset Demonstrated 80% accuracy on the multi class classification problem with a single Convolution layer and max pool layer