Multi-source Machine Learning for Computer Vision

Vision

We focus on self-supervised and human-machine collaborative learning of robust representations using contrastive, zero-shot, multi-modal and transfer learning.

Funded Projects

• A data-driven multi-scale phytotechnology framework for identification and remediation of leached-metals-contaminated soil near coal ash impoundments – DOE 2022
• Reducing Operating Room Waste by Monitoring Single-use Sterile Surgical Supplies with Computer Vision – NIH iTHRIV 2021
• Determination of Safety Limits against Cyber Threats in Neuromodulation Devices using Machine Learning, Brain Phantoms, and Neural Pathways – Commonwealth Cyber Initiative (CCI) 2021
• Intelligent Augmented Reality for the Future of Work – VT ICTAS EFO 2021
• Intelligent Augmented Reality for the Future of Work – VT CHCI 2021
• Pull-Off Arm and Vibration Stoppers Detection – Japan Rail East Information Systems 2018
• Augmented Reality Assistant – Xerox 2017
• Automating Evaluation of Product Efficacy – Procter and Gamble 2017
• Connected Consumer for Enhancing Product Experience – Procter and Gamble 2016

Publications

• System and method for performing collaborative learning of machine representations for a target concept – Patent Published 2022
• System and method using augmented reality for efficient collection of training data for machine learning – Patent Granted 2022
• Agile video query using ensembles of deep neural networks – Patent Granted 2021
• Building Jarvis-A Learner-Aware Conversational Trainer – IUI Workshop 2019
• Hybrid image-based defect detection for railroad maintenance – Electronic Imaging 2019
• Hard Negative Sampling Strategies for Contrastive Representation Learning – Preprint 2022