Finding the geoposition of objects from sensory data is important for higher-level inference. We have developed different techniques using deep learning and Bayesian inference that works on images, inertial measurements, lidar, and their integration. Please visit the project page for projects.
Visual Object Tracking
Detection and tracking objects in images and image sequences can be considered a mid-level computer vision problem. The goal of the research we conduct under this category relates to solving the spatial and temporal problems related to funding the objects and generating consistent trajectories. Read more on the project page.
We have worked on a number of medical imaging projects related diagnosis of patient conditions from lung ultrasound, CT scans, and Gamma imaging. Our techniques range from deep learning approaches to more traditional machine learning methods. See here for more information.
Big Data Analytics
Semantic Scene Labeling