With all of the talk about machine vision in the Artificial Intelligence and Machine Learning world, one thing is quite evident; it can end up consuming large amounts of power and space if you are not careful what platforms you chose. If power consumption is a key factor along with space constraints, then you might just want to look at the Microchip PolarFire FPGAs and VectorBlox platform.
PolarFire FPGAs consume far less total power than competitive mid-range FPGAs -- 30-50% lower, actually -- which makes them ideal for edge devices that require intensive data analysis while keeping power usage at a minimum. This puts them into a category for utilization in applications from video to imaging and many machine learning applications with high performance requirements in small form factors. Lowering the power usage can also eliminate the need for additional heat sinks and thermal fans further reducing overall package size.
Check out the PolarFire Video Kit (MPF300-VIDEO-KIT-NS) for a seamless development package to test and design your next embedded vision product or system. In addition to the hardware platform is, of course, VectorBlox, a full accelerator SDK for compiling a neural network from TensorFlow and ONNX into Binary Large Object (BLOB) that will be stored in flash and loaded to the memory during execution. VectorBlox allows for easier porting and development of neural networks into FPGAs to give designers the tools and resources needed, even without extensive experience.
So, when you need an optimized solution for your edge device that brings down development time, power consumption, and form factor, Microchip’s PolarFire FPGAs are certainly a platform to investigate. Take a look at the full family of parts and wide array of documentation support that will bring an added level of intelligence to your part of the design space. When you couple that with the vast portfolio of components, working with Microchip is an easy decision.
