I am interested in enabling machines to "see", and answer questions about the scene or manipulate the captured media for creative effects. My research is in the area of computer vision and computational photography. I've worked on various novel camera systems and algorithms for scene understanding and cinematic effects.
At Intel labs I have been working on various multi-camera and light field capture systems & algorithms. My team worked on the Intel RealSense snapshot technology from research to prototype to helping the business unit productize as part of Dell Iconic tablets. I delivered the RGB-D segmentation and accurate 3-D measurement algorithms. My intern developed background cleaning and fusion algorithms for these cameras. I've put together multiple cinematic experiences and applications using the algorithms. Currently, I am exploring traditional and deep learning based algorithms for multi-camera, light fields and other novel camera systems towards creative editing effects and immersive displays.
My PhD research was focused on scene understanding using RGB and depth data from various passive, active or hybrid camera systems. Using both texture and structure information allowed for algorithms that could answers questions about the scenes and objects, such as
What objects do I see in this scene? Which objects do I already know? Can I update my knowledge about the objects I knew and learn new ones? Is this an object I can sit on? Which surface of this object is useful for sitting? Where can I place things on, on this object? I had only seen chairs before, but I can recognize that a sofa is also Sittable. Can I add that to my canonical model of sittable class?Where am I? Is this a kitchen or living room? Where am I? Which place is it?
I've worked on various stereo & hybrid multi-camera systems combining different resolutions, active depth sensors, flash based projectors.
I created a dataset, VADANA, to study algorithms that can verify if two people could be siblings, related by blood or not. Its also used for problems of face verification, gender & age estimations.