Bruno Lavigueur, Embedded Vision Subsystem Project Leader at Synopsys, presents the "Combining Flexibility and Low-Power in Embedded Vision Subsystems: An Application to Pedestrian Detection" tutorial at the May 2014 Embedded Vision Summit.
Lavigueur presents an embedded-mapping and refinement case study of a pedestrian detection application. Starting from a high-level functional description in OpenCV, he decomposes and maps the application onto a heterogeneous parallel platform consisting of a high-performance control processor and application-specific instruction-set processors (ASIPs). This application makes use of the HOG (Histogram of Oriented Gradients) algorithm.
Lavigueur reviews the computation requirements of the different kernels of the HOG algorithm, and presents possible mapping options onto the control processor and ASIPs. He also presents an OpenCV-to-ASIP software refinement methodology and supporting tools. He presents detailed results of the final configuration consisting of one control processor and four ASIPs, including cost and power figures. Finally, he summarizes the results on an FPGA-based rapid prototyping platform.