Image Processing
Face Cropper (Commercial Project)
This is part of a commercial project that finds a face in an image and crops as symmetrically as possible to fit on ID cards .
Remote Sensing
Map Sentinel-2 with SOC Ground truths
Programmatically created region of interest (ROI) and associated Sentinel-2 Multispectral data with the measured Soil Organic Carbon (SOC) ground truths.
PyTorch Projects
Hyperspectral image segmentation on the Indian Pines dataset
Using 3D convolutional neural network, developed a model that detects class of a hyperspectral image patch.
Multi-task learning
From a multi-labelled face dataset, predict age and gender.
Speech emotion recogniser
Recognises emotion based on SAVEE dataset using CNN.
Currently, working on using transformer network to capture temporal relation for better accuracy.
Understanding the blackbox by exploring saliency map
Saliency map highlights the pixels (in computer vision), which are responsible in making a decision. This projects generates saliency map on an MNIST digit classifier.
Detect glasses using transfer learning
Checks if the person in the image is wearing glasses. Used Resnet-18 as backbone.
Soil moisture prediction using multimodal data
Predicting soil moisture from time-series data from different sensors including hyperspectral camera.
Stock prediction using LSTM
Predicting Amazon stock price using LSTM.
Chronic Kidney Disease detection
Detects CKD with 98% accuracy.
Using partial derivative method, depicts impact of each feature in decision making process of the machine.
Estimating nitrous oxide emissions
Simple NN regression model to estimate N2O emissions
Autoencoder
Simple Autoencoder on MNIST dataset
Variational Autoencoder
Simple VAE on MNIST dataset. Used tensorboard to visualise continuous progress.
Speech command recogniser
Simple FC network to recognise short speech commands (for example, forward, up, yes, no etc.).
A deterministic machine with predefined weight initialization
An experimental work in search of explainable and deterministic neural networks.
First started with creating 28X28 filters for each classes (digits in this case) by accumulating the average distribution of the pixels.
Without any backpropagation, it works with 77% accuracy. Improving the filters with backpropagation (not randomly initialized) produces 88%+ accuracy.
Future work in this would be applying DSC and addressing the issue of lack of translational invariance.
Python Projects from Scratch
CNN to recognise speaker
CNN built from scratch to recognise speaker from mel spectrogram data.
Single Layer Network
A simple neural network from scratch for demonstration purpose
Polynomial regression
Aligning a polynomial function to sine curve for demonstration purpose
SnakeGame with PyGame
Classic snake game. A foundation work to learn reinforcement learning.
Conway's Game of Life with PyGame
Simulate Conway's Game of Life. Auto zoom-out when generations exceed borders.
Unity
Quad
Implementing Quad (Samuel Beckett) by using dynamic mesh generation. Here is a quick demo.
Bioinformatics
Rosalind Solution
An attempt to learn fundamental of genome sequencing to be able to apply machine learning.
Works in Progress
AI - Escape game (Deep Q Learning)
In this game AI agent has to escape through the hole avoiding enemies.
Done: Basic gameplay
Remaining: AI Agent
3D Image segmentation using U-Net
Spotting tumor cells on brain using the BraTS dataset
Done so far: Processing nii files. Prerequisite knowledge - U-NET, 3D Convolution, DSC similarity etc.
New faces using GAN
Using CelebA dataset, generating new faces.
Done so far: Initial codes and tensorboard setup for debug and diagnosis.
Articles
Just another Neural Network from scratch with mathematical intuition
Feel calculus with your programming sense
Normal Distribution — an intuitive introduction without math (paid work)
Single Neuron Neural Network on Python — with mathematical intuition (paid work)
An intuitive explanation of how meaningless filters in CNN take meaningful shapes
One of the millions of learning paths to deep learning for programmers
Professional Projects (2007-2020)
- Enterprise billing system comprising 10+ modules using Grails, MySQL, jQuery (autobill.com)
- A robust, and user-friendly CMS using Grails, MySQL, jQuery (webcommander.com)
- Refurbished one of the pioneer CMS in Australia using Java Swing, CakePHP (webconsole.co)
- Application deployment and management system using Java, PHP, Python, Bash scripting
- Payment processing service using CakePHP and MySQL
- Customized automated test suites and CI environment with Grails and Selenium
- A video processing system which takes a video of a graduation ceremony as input, and generates short clips for each students based on the speeches in the video using Python and Laravel
- Contributed to hypercasual mobile games using Unity3D (themascoteers.com/games)
Projects Done During Undergraduate (2004-2007)
- Puzzle solver in C using graph searching algorithms
- Parallel processing system using Java RMI
- Web-based expert quiz system using CakePHP and MySQL
- Complete English-Bengali dictionaly using Java Swing
- Library management system using .NET
- Video club management system using JSP, Servlet, jQuery
- DOM-based XML parser using Expat, an event-based XML parser on C
- File transfer system over the loal network using socket programming on C
- Drawing natural objects using OpenGL on C
- Console-based abstract strategy game using minimax algorithm on C++