Natasha Pandit

Papers

Total Citations

H-Index

About

Natasha Pandit is a researcher at the forefront of fine-grained action recognition, specializing in the intersection of computer vision, machine learning, and smart health. Her work addresses a critical gap in the field: the identification of rapid, sub-second actions—a challenge far removed from traditional coarse action classification like running or climbing. Pandit’s major contribution is the development of the **StrokeRehab benchmark dataset**, a pioneering resource designed specifically for high-temporal-resolution action identification from video and kinematic data. This dataset, introduced in her 2022 paper, has already garnered **5 citations**, signaling its growing importance as a standard for the community. Complementing this, her 2021 work on **sequence-to-sequence modeling** for high-resolution action identification further advances the methodology, enabling more precise analysis of rapid movements. These contributions have profound implications for **robotics, rehabilitation, and smart health**, where understanding subtle, fast motions is essential. Pandit’s research is particularly notable for its focus on **stroke rehabilitation**, offering tools that could transform patient monitoring and therapy. By pushing the boundaries of temporal resolution, she is enabling machines to see and interpret human motion with unprecedented granularity.