SURF cascade face detection acceleration on Sandy Bridge processor

Along with the inclusion of GPU cores within the same CPU die, the performance of Intel's processor-graphics has been significantly improved over earlier generation of integrated graphics. This paper presents a highly optimized SURF cascade based face detector which efficiently exploits both CPU and GPU computing power on the latest Sandy Bridge processor. The SURF cascade classifier procedure is partitioned into two phases in order to leverage both thread level and data level parallelism in the GPU. The integral image function running in the CPU core can work with the GPU in parallel. We measure the performance and power of the GPU implementation on the latest Sandy Bridge platform. The experimental results show that our proposed GPU implementation achieves a 2.98 speedup and a 1.42 speedup compared to the single thread and multi-thread CPU implementation. At the same time, the power usage can be reduced as much as 50% compared to the CPU implementation. In addition, our proposed method presents a general approach for task partitioning between CPU and GPU, thus being beneficial not only for face detection but also for other computer vision applications.