The video below, made at the recent Embedded Vision Summit, demonstrates an Intel® FPGA performing face recognition through a machine language (ML) inference model. A CPU downloads the model to the FPGA in encrypted form and is stored that way in the FPGA’s external SDRAM. Decryption only occurs when the FPGA reads the model and stores it in the on-chip SRAM, thus protecting the IP associated with the face-detection IP.
For more information about Intel Vision products, click here.
You might also want to download the White Paper titled “Unleashing the Power of Intel® Vision Products for Facial Recognition using Deep Learning Inference Acceleration.”