"Urine Sediment Analysis"
We developed an urine sediment analyzer using microscopic urine sediment images. The development involved various techniques from Deep Learning, Computer Vision, CNN & GAN for Arkray Inc. Japan. Further details are prohibited under Non-Disclosure Agreement.
This work produced a Research Publication at MICCAI 2019 & a Patent (filed patent in japan).
"Semantic Multinomial Representation for Scene Images using Pseudo-concepts and Concept Neural Network"
In order to improve scene image classification we formulated Semantic multinomial(SMN) representation for images by deriving pseudo-concepts from Convolutional Neural Networks. We developed Concept neural network (CoNN) to classify these SMN representations.
"Odia Language Stemmer"
We developed a stemmer for Odia language during my B.Tech at ITER, Odisha (India). A task opted from FIRE-2012, Indian Statistical Institute Kolkata. A suffix stripper algorithm was implemented (without using YASS tool), with some limitations.
This course project aims to classify different patterns of signals (e.g, Images, Speeches, Data points etc.). The project was meant to address to classify linear, non-linear and overlapping data.
It involved machine learning techniques like Baye's classification, GMM, KNN, SVM, NN etc.
The project was developed using MATLAB.