Table of Contents
Research Projects
The Impact of False Positives on the Human Trustworthiness of Deepfake Detection Algorithms
In this project, my team and I are researching the trustworthiness of humans on deepfake detection algorithms when the performance of the machine with varying false positivity rates is reveiled to the users. The two hypotheses being tested in this research are: If false positive errors will decrease human trust in deep fake detection algorithms and if the trust will mediate the decision making with false positive errors and deep fake detection algorithms.
Mentors: Dr. Aiping Xiong & Dr. Sarah Rajtmajer- College of IST, Penn State University
Human Emotion Recognition using EEG through Music-Listening
In this project, the objective was to classify EEG data, generated through music-listening, into discrete emotions. I developed a hybrid feature-extraction model by using three domains of an EEG signal: Time, Frequence and Time- Frequency Domains and implemented CNN model as well as state-of-the-art ML algorithms such as XGBoost and Random Forest.
This project won first runner up in IEEE GRSS Women's PG Thesis Competition 2021 as well as was accepted in IEEE Indicon 2021 conference.
Mentor: Dr. Anup Nandy, Assistant Professor-CSE dept, NIT Rourkela
Dynamic Hand Gesture Based Password Recognition System using Webcam
In the project, I created a dynamic hand gesture based password recognition system using neural networks. The uniqueness of this system lies in the fact that it would use an ordinary low-cost webcam (built into the laptop) rather than a Kinect camera so it can be implemented into everyday electronics and add an extra layer of security. I worked on Octave to complete this project.
This project was accepted in the ICCIS 2019 conference and is published in the Springer LNNS journal.
Mentor: Dr. Arun Kumar, Assistant Professor-CSE dept, NIT Rourkela
Development of a multi-modal gait analysis tool to estimate the human cognitive state
In this project, based on computer vision and machine learning techniques, A multi-modal gait acquisition system is under development using Microsoft Kinect sensor and IMU motion sensors for capturing the human gait patterns on different environmental contexts and to analyze the subject’s cognitive reactions on multiple mental states. In this manner, the cognitive state can be predicted by observing various gait patterns. My role in this project was to implement a multi (8 Cameras) calibration pioneer system in such projects.
This project was accepted in the CVIP 2021 conference and will be published in Springer CCIS journal.
Mentor: Dr. Anup Nandy, Assistant Professor-CSE dept, NIT Rourkela
Facial Detection and Tracking using Viola Jones
This project was a combination of hardware and software programming. The facial detection was done using the Viola Jones Algorithm after which coordinates were extracted from the detected region and sent to an arduino. A web cam mounted on two servos for tilting and panning, was controlled by the arduino to track the detected face.
Mentor: Dr. Sujata Mohanty, Assistant Professor-CSE dept, NIT Rourkela