MY PROJECTS

EchidnaLB

EchidnaLB

Software Development
EchidnaLB is a high-performance Layer 7 load balancer written in Rust, designed to manage HTTP/HTTPS traffic with support for HTTP/1.1 and HTTP/2 protocols. It offers multiple load-balancing algorithms, including Round Robin, Weighted Round Robin, IP Hashing, Least Connections, and Least Latency, ensuring efficient traffic distribution across backend servers. The load balancer supports both IPv4 and IPv6 listeners, configurable health checks to monitor server availability, and optional TLS termination with SSL certificates for secure communications.
Iptables Visualizer

Iptables Visualizer

Kubernetes, Linux
A web application for visualizing the iptables rules present in the Kubernetes kube-proxy pods or in a Linux operating system in the form of a directed acyclic graph.
Grid Path Finder

Grid Path Finder

Algorithms, Data Structures, Web Development
A web app for finding and visualizing the path between a source position and a destination position in a grid using graph algorithms such as BFS, DFS, Dijkstra, etc.
Chess AI

Chess AI

Artificial Intelligence, Web Development
An AI chess engine developed in Javascript that implements a Mini-Max algorithm to perform an adversarial search with the objective of finding the next best move. To make the process of searching the game tree more efficient, Alpha-Beta pruning is incorporated in the logic. The heuristic function to evaluate the board position is defined as the sum of individual piece cost weighted according to the piece-square table. Finally, there's a leaderboard for those who successfully beat the AI, which is implemented using Firebase Cloud Firestore.
Sorting Visualizer

Sorting Visualizer

Algorithms, Web Development
A web application for visualizing Bubble Sort, Insertion Sort, Selection Sort, Merge Sort, and Quick Sort. The application is developed in next.js application developed along with react.js for server-side rendering of react js application for improved SEO and support for hardware resources constrained devices. The codebase is written in Typescript instead of Javascript for better code understandability and maintainability.
Image Forgery Detection

Image Forgery Detection

Deep Learning
The advent of social media has leady to a steadily increasing cases of fake news which seek to promote communal violence, fake scandals, etc. to name a few. The implemented Deep Learning model called ManTraNet (based on 2019 paper of Wu et. al.) is able to detect 385 different types of image forgery and localise the forged region as well.
Neural Style Transfer

Neural Style Transfer

Deep Learning
A neural style transfer algorithm which uses a style image to redraw a content image like the style image.
Driver Drowsiness Detection

Driver Drowsiness Detection

Deep Learning, Computer Vision
A real time driver drowsiness detection system which alerts the driver if they fall asleep due to fatigue. The computer vision software is able to detect forward head tilt angle, closing of eyes and yawning.
Parkinsons AI

Parkinsons AI

Deep Learning, Mobile App Development
An AI based android app which is able to diagnose the Parkinson's Disease using two independent tests that require only a pencil and a paper. Based on 2017 research paper Distinguishing Different Stages of Parkinson's Disease Using Composite Index of Speed and Pen-Pressure of Sketching a Spiral by Zham et. al.
Crop AI

Crop AI

Deep Learning, Mobile App Development
A diagnostic AI aided mobile app which is able to classify up to 38 different plant diseases ranging for 14 crops and vegetables.
ASL Sign Language Translator

ASL Sign Language Translator

Deep Learning, Object Detection
An object detection model that uses an artificial neural network to label the ASL hand sign alphabets. This deep learning model makes use of the Faster RCNN model architecture to perform the object detection.
Blind AI Assistance: Vision

Blind AI Assistance: Vision

Deep Learning, Mobile App Development
Assisting blind people with the help of AI aided image captioning via a smartphone app.
Colorize: Color B/W Images using AI

Colorize: Color B/W Images using AI

Deep Learning
Colorize is a computer vision algorithm which is implemented as a feed-forward pass in a CNN at test time, that has been trained on a million images from the Imagenet dataset. The algorithm is able to fill vibrant and realistic colors to the black and white images to recreate a plausible colorized image. Based on the 2016 research paper by Zhang et al. "Colorful Image Colorization".
YOLOv3 Object Detection

YOLOv3 Object Detection

Deep Learning, Object Detection
An implementation of YOLO object detection deep learning algorithm, trained using Microsoft's COCO dataset and coded using TensorFlow framework for ML.
Sudoku Solver

Sudoku Solver

Python, Algorithms
A simple Python based Sudoku Solver based on the backtracking algorithm that has been implemented along with a GUI, made using Pygame library.

CONTACT ME

© Neelanjan Manna, 2020 - 2024