Synthesizing Machine Learning Training Data from Privacy Preserving Video
June 4, 2024Inspiren Computer Science/Mathematics, 2023–24
Liaison(s): Kyle Lampe ’02, Bryce Chudomelka
Advisor(s): Gabe Hope
Students(s): Zoe Shao (TL-F), Mehek Mehra (TL-S), Nicole Falicov (TL-S), Aidan Stone-Grijalva
Inspiren is a healthcare technology company dedicated to assisting caregivers through the use of AI. Their flagship product AUGi reduces risks to patients by providing additional monitoring through computer vision techniques, ensuring caretakers are able to optimally visit patients. Due to privacy concerns, the biggest challenge in training AUGi has been finding robust quantities of training data from healthcare settings. The Inspiren Clinic team has evaluated different ways of generating training data, including synthesis and anonymization techniques.