CRICS - A Cricket Scene Dataset & Classifier

Created a large dataset from 55 match highlights, refined through Denoising Autoencoder-based semantic Hashing to eliminate redundancy and enhance scene detector model's generalizability

Event Detection in Cricket

Using cricket highlights from YouTube trained state-of-the-art CNN models to accurately detect ball, batsman, bowler, bounce, shot-direction and hit (using the Mel-spectrogram of the audio).

Emperically Proving the Monty Hall Problem

Created a simulation of The Monty Hall Problem and showed that switching from the original decision always results in a higher number of victories in the case of 3 doors.

Anime Face Generator Using DC-GAN & VAE

Emotion Detection Using Audio

Sentiment Analysis Of Amazon Fine Food Reviews

Extensive Study Of T-Sne For Visualization

Highlights Of Developed Algorithms

‒ Multithreaded Matrix Multiplication (Java) | Achieved 70% speedup in large matrix multiplication.
‒ External Sorting for large files (C) | Achieved a speed of 3MB/second on i7 7th gen.
‒ Classical Knight Tour and N Queen Problem in chess using backtracking (C).