I work broadly in the field of Computational Imaging — which involves co-designing optics, sensors, and algorithms to perform novel sensing tasks. [Google Scholar]
Projects (Current)
Computer Vision for Surgical Tool Tracking
Using the latest advancements in computer vision for tracking surgical tool-tips and other keypoints. We want this tracking to be robust to occlusions/motion blur, and should work on long videos. Such an automated tracking method can potentially enable downstream video analytics, such as surgeon motor skill assessment, learning assessment, etc.
Split Aperture Coded Computational Photography
Coded photography has enabled 3D estimation, depth sensing, hyperspectral sensing, etc. I have explored solutions in which a single camera can capture multiple coded images at the same time, leading to better 3D estimation/depth sensing/deblurring. This snapshot multi-coded captures can be achieved by intelligently capturing coded measurements in different dimensions (such as polarization, dual-pixel measurements, etc.). See my ICCP 2022 and CVPR 2024 works.
Laser Speckle Imaging for Deep Artery Blood Flow Sensing
Designing a wearable device and speckle estimation algorithms for sensing blood flow changes in deep-lying peripheral arteries (5-10mm deep).
Projects (Past)
Coded Fiber Optics Imaging
Attempted to use phase masks for 3D estimation or multi-depth imaging using a fiber bundle-based flexible imaging system (HRME).
Depth from Defocus + Shape from Polarization
Attempted to merge depth from defocus cues with polarimetric signal cues that give object shapes.
2D Phase Unwrapping
Designed optimization-based 2D phase unwrapping algorithms. See my NCC 2018 and TGRS 2021 works.