Research and Awards
In the last few years, I have been fortunate enough to be mentored by some of the most eminent researchers and work with awesome team members. This page has insights into the work that I have been doing and the appreciation that work has received from the (generous) community.
Projects
MuteIt: Jaw Motion-Based Unvoiced Command Recognition Using Earable
MuteIt is an ear-worn system for recognizing unvoiced human commands. MuteIt presents an intuitive alternative to voice-based interactions that can be unreliable in noisy environments, disruptive to those around us, and compromise our privacy. It combines IMU data with the anatomy of jaw movement as well as principles from linguistics, to model the task of word recognition as an estimation problem. Validated for 20 subjects with diverse speech accents over 100 common command words, MuteIt achieves a mean word recognition accuracy of 94.8%.
SpiroMask: Measuring Lung Function Using Consumer-Grade Masks
SpiroMask retrofits a microphone in consumer-grade masks (N95 and cloth masks) for continuous lung health monitoring. Our approach is evaluated on 48 participants (including 14 with lung health issues) and finds that we can estimate parameters such as lung volume and respiration rate within the approved error range by the American Thoracic Society (ATS).
[Paper to be updated after camera ready]
JawSense: Recognizing Unvoiced Sound using a Low-cost Ear-worn System
JawSense enables a novel form of human-computer interaction based on unvoiced jaw movement tracking. JawSense allows its user to interact with computing machines just by moving their jaw without producing any sounds. JawSense can be used as a form of hands-free and privacy-preserving human-computer interaction with a 92% phoneme classification rate.
Given the pandemic and the high air pollution in large parts of the world, masks have become ubiquitous. We envision masks to monitor health-related parameters such as lung activity, use retrofitting sensors, and display localized pollution. This can create awareness about air pollution. In this work in progress, we develop Naqaab, a mask that measures the forced vital capacity (FVC) of the lung using a retrofitted microphone. The results are displayed on a mobile application along with pollutant levels measured using the microphone.
Sentiment analysis of unproven strategies
For mitigating the rising levels of air pollution in India, policymakers are now turning to unproven strategies that provide a false sense of security. We analyze the sentiment around such policies and schemes using tweets. Also, we identify the key players that derive the discussions around unwarranted policies.
Survival Analysis: Multiple Myeloma
Predicting the stage of Multiple Myeloma (a type of blood cancer) accurately requires expensive lab tests and a lot of data collection. We propose a staging system based on inexpensive blood tests incorporating subject features like age, gender, commodities, etc.
Air Cognizer
Air Cognizer is an android application that predicts air quality with smartphone camera photos, enabling users to get real-time, localized, and accurate Air Quality Index (AQI). It allows users to upload an image of the sky horizon to get AQI with corresponding health warnings.
The project was mentored by Dr. Aakanksha Chowdhery (Google AI) and Dr. Brejesh Lall (IIT Delhi).
[Video] [Website] [Playstore] [Poster] [Marconi Society] [BusinesWire ]
Prosthetic Arm
The prosthetic arm is an artificial device that is capable of replacing a missing or dysfunctional intended to restore the normal functioning of the upper limb. The project involves the development of a low-power electronic circuit board and software for the prosthetic hand using Myo Electrodes for EMG. This project is developed for Endolite India Ltd.
Music instrument and note detection
The audio signal was used for identifying the instrument it was played on and the note of the same. Constant Q transform was applied to the input signal which eventually served as input to a 3D CNN.
Awards
Graduate Research Day Stony Brook University - Best Presentation Award 2022
The Graduate Research Day is the revival of the Graduate Research Conference. All graduate students in the Department of Computer Science are invited to submit an abstract and poster of their research work and virtually attend the event. MuteIt was awarded the best presentation award across 36 submissions from different domains of systems, security, and machine learning, The grading criteria were based on novelty, execution, presentation, and overall impact.
Graduate Research Day Stony Brook University - Social Impact Award 2021
The Graduate Research Day is the revival of the Graduate Research Conference. All graduate students in the Department of Computer Science are invited to submit an abstract and poster of their research work and virtually attend the event. JawSense was awarded the social impact award as the system can be used by people with speech disorders to interact with smart devices.
Paul Baran Young Scholars Celestini Prize India 2018
The vision of the Celestini Program is to solve an important socio-economic problem (such as road safety, air pollution, and water quality) using machine learning and data analytics on a network of IoT or mobile devices. It is funded by the Marconi Society which is a foundation dedicated to benefiting mankind through scientific achievements in the internet and communications. More than one hundred students expressed interest in working on the program during the summer and three student teams (comprising eight students) were selected. My team worked on Air Pollution estimation and created Air Cognizer. The team was awarded $1500 along with a stipend of $200 for each team member.
Evotech is a project exhibition launched with a vision to let young developers portray their products in solving real-world problems. The teams are judged on 3 criteria namely innovation, feasibility, and reachability of the project.