Research
Graduate Research Assistant | SVCL Lab
Location: UC San Diego, CA, USA
Duration: June 2023 - Present | 2 mos
Responsibilites
- Working on Accurate 3D-Hand Pose Estimation via Multi-Modal Fusion
- Researching multi-modal fusion techniques to improve 3D hand pose estimation in highly occluded scenarios. Developed a synchronized and automated approach to collect a well structured dataset from multi-view cameras and a wearable sensor.
- Conducted extensive experiments with the Meshgraphormer model on the developed dataset, exploring various modalities. Identified the need for additional data collection to improve 3D hand pose estimation in occluded scenarios.
Graduate Research Assistant | Dr.Yuanyuan's Lab
Location: UC San Diego, CA, USA
Duration: Dec 2022 - Apr 2023 | 5 mos
Responsibilites
- Worked on Data-driven Smart Building Modeling
- Developed an efficient data storage system using PostgreSQL to organize and store temporal data, improving data access speed by automating table partitioning based on sensor IDs.
- Developed Python Flask APIs that connect to the database and retrieve data for a given time interval and registered users, while also developing a ReactJS-based frontend website with pages for registration and data retrieval.
- Performed exploratory data analysis on sensor data, extracting vital features. Implemented linear regression and random forest for accurate feature prediction, establishing a baseline.
Undergraduate Researcher
Mentor: Dr. Lakshmi Narasimhan
Location: IIT Palakkad, Kerala, India
Duration: Jan 2020 - Apr 2020 | 4 mos
Responsibilites
- Worked on developing Software Defined Radio for Visible Light Communication
- Developed an Open SDR to aid the research in visible light communication. The open source SDR invites people to experiment with the setup. We developed software using Python Tkinter and matplotlib to make a GUI for SDR and implemented a hardware model using LEDs and phototransistors.
- We were able to transfer data but with very low bit rates. Hence, we designed a more efficient model to increase the bit rates but couldn’t implement it because of Covid-19 restrictions.