sed in those operations that can be parallelized. The CUDA compilation flow can be seen in Figure 3. 3. Hardware for detecting and tracking vehicles in ComViDAS In terms of computing power, The Jetson TK1 development kit from Nvidia can deliver up to 326 GFLOPS. This metric exceeds any of the single-board computers studied so far. This and previous arguments led to the selection of the Jetson TK1 Development Kit as the best available embedded solution for the ComViDAS project. As the Jetson TK1 supports CUDA 6.5, it can be used to handle the CPU and GPU at the same time . The CPU in this project will be used in sequential tasks while the GPU will be u