I-LAIST simultaneously tracks and detects moving targets in large-scale visual data, such as wide area motion imagery (WAMI). Existing visual tracking algorithms can only track a limited number of targets in a small area.

I-LASIT operates within the framework of cloud computation and Graphic Processing Units (GPUs). It uses a front end and a web-based server to call Hadoop and performs highly parallelized computation functions based on CUDA/C++. This computational architecture processes images at a real-time rate. I-LAIST provides drastically improved tracking with low frame rates over realistic conditions. The resulting trajectories can be uploaded into IFT’s Trajectory Analysis Program (I-TAP) to perform Pattern of Life (PoL) analysis on the traffic flow.