Won first prize at the Tracking and Imaging Challenge at PCV-2014 Conference

This Challenge was to use both Trajectory data from sensors and Image data for solving some unorthodox and novel applications. It was held by ISPRS (Intln. Society of Photogrammetry and Remote Sensing) as part of the Commission III - Photogrammetric Computer Vision Symposium in Conjunction with the European Conference on Computer Vision at Zurich (5-7 Sept 2014) and supported by ACM SIGSPATIAL.
The paper was "Deepak Rajamohan, Bhavana Ganu, K. S. Rajan: Road Condition and Texture Estimation by Fusing GPS, Accelerometer and Camera Data"

The proposed idea was to use GPS, Accelerometer and Image data (all mounted on a car dashboard) to identify the road surface types, potholes, etc as a low-cost alternative to the rather expensive RoadScanning Systems (also known as Real-Time profiler).

You can see the prototype working at http://www.youtube.com/watch?v=geu6IoKTqw4

More details at http://www2.isprs.org/commissions/comm2/wg8/tic.html