Follow Us On

Sub-500mW facial recognition solution on Microchip’s PolarFire FPGA using the VectorBlox AI Accelerator SDK

October 14, 20219:30 - 10:00AI Hardware - AI Expo 2021


David Wyatt

David Wyatt

CTO, President, PixelDisplay
  • October 14, 2021
  • 9:30 AM - 10:00 AM


Diptesh Nandi

Diptesh Nandi

Sr. Product Marketing Manager, Industrial and Automotive, Microchip Technology Inc.


VectorBlox is an SDK that enables ML inferencing on our PolarFire FPGAs. The SDK has tools to convert typical neural network models which can be run without requiring re-programming of the FPGA. As PolarFire is the most power efficient mid-range FPGA in the industry today, VectorBlox can enable the power efficient inferencing solutions, running on 2-3 times less power compared to competitive products, for a similar computational load.

Other leading-edge solutions also available, include a dual pipelined CNN based facial recognition solution without requiring any FPGA programming. Accelerating and reducing development cost in well defined use cases such as intruder detection, with military/commercial or at-home surveillance customers.