We are interested in building practical systems that can leverage latest technologies to benefit society.
Traffic
AutoCalib is a system for scalable, automatic calibration of traffic cameras. AutoCalib uses deep learning to extract selected key-point features from car images in the video and then uses a novel filtering and aggregation algorithm to automatically produce a robust estimate of the camera calibration parameters from just hundreds of samples. We have implemented AutoCalib as a service on Azure that takes in a video segment and outputs the camera calibration parameters. Using video from real-world traffic cameras, we show that AutoCalib is able to estimate real-world distances with an error of less than 12%.
The obtained camera calibration can be used to build interesting smart-city applications. For example, one can detect speeding cars from the video feeds automatically. For more details, refer to our ACM BuildSys 2018 paper (best paper award and best demo award winner).
Pollution
Pollution is plaguing many of our cities today but its cause is multi-factorial and it is not clear what we can do about it. A first step towards tackling pollution is understanding how pollution varies at a geographic location in a fine-grained manner (e.g., near us, where we travel, etc. rather than at city level). We show that pollution can be measured scalably in a fine-grained meanner. Our measurements show that pollution at micro-scale exhibits interesting properties that can perhaps indicate ways for us to minimize impact of pollution on us. See our paper at COMSNETS 2018 WACI workshop (best paper award winner).