My 2013–2014 undergraduate project ran an OpenCV face-detection application on the Altera SoCKit development board. The board combines an ARM processor subsystem with FPGA fabric. The work covered board configuration, Linux, USB camera support, and the application software.
Board and software integration
QSys and Quartus were used for the board configuration. The Linux build used the Linaro toolchain, with USB Video Class support enabled for the webcam. OpenCV provided the image-processing library used by the application.
The camera, kernel, libraries, and application formed one processing path. A usable camera image depended on the device being recognized by Linux and the application receiving it in the expected format. The project archive contains the thesis, development files, and demonstrations of board startup and camera operation.
Performance comparison
The thesis compared the SoCKit application with an Intel Core i7-2670QM system using a standard Lena image. It reported the SoC implementation as approximately 43% slower for that setup and workload.
The measured implementation established a software baseline. The conclusions proposed FPGA acceleration as subsequent work. Evaluating that extension would require selecting an operation to move into programmable logic and measuring its processing, data-transfer, and synchronization costs.
Publication
The research was carried out in 2013–2014. The related paper, “Integration of OpenCV and Cyclone V Hybrid ARM and FPGA SoC for Face Detection Application,” was published in the ISCIT 2023 proceedings. I am its first author, with Jefferson A. Hora, Gene Fe P. Palencia, Harreez V. Quimque, and Nieva M. Mapula.