Microchip Announces Smart Embedded Vision Initiative for Machine Vision System Design
Using Microsemi PolarFire FPGAs, the new program aims to enable compact, high-bandwidth, low-power system design.
Using Microsemi PolarFire FPGAs, the new program aims to enable compact, high-bandwidth, low-power system design.
Today’s compute-intensive, vision-based systems require high-bandwidth network communication. At the same time, power requirements are tight and space is at a premium.
To meet these challenges, Microchip has introduced its Smart Embedded Vision Initiative. The new ecosystem leverages Microchip’s low-power PolarFire FPGAs with a series of high-speed imaging interfaces, intellectual property (IP) for image processing and an enhanced ecosystem of outside partnerships.
The Microchip PolarFire FPGA. Image from Microsemi
The initiative is designed to bolster machine vision advances in many applications that increasingly utilize vision processing:
- Industrial
- Medical devices
- Broadcasting
- Automotive
- Aerospace and defense
- Cellular infrastructure
Polar Fire FPGAs
Members of the MPFx00T PolarFire FPGA product family include the range from 100K logic elements (LEs) to 500K LEs. They feature 12.7G transceivers and use up to 50% less power than competing mid-range FPGAs. You can learn more about the family here.
Smart Embedded Vision IP
In January, Microsemi released a new version of their SoC FPGA dev platform, the Libero SoC Design Suite. The IP tools and cores announced this week are available through this suite.
The new initiative includes, but is not limited to:
- Serial Digital Interface (SDI) IP: Used to transport uncompressed video data streams over coaxial cabling, this interface supports multiple speeds, including: 1.485 Gbps, 2.970 Gbps, 5.94 11.88 Gbps.
- MIPI-CSI-2 IP: A sensor interface often used to link industrial cameras to FPGAs. The PolarFire family supports receive speeds up to 1.5 Gbps per lane and transmit speeds up to 1 Gbps per lane.
- 2.3 Gbps per lane SLVS-EC Rx: An image sensor interface IP that supports high-resolution cameras. Two-lane or eight-lane SLVS-EC Rx FPGA cores can be implemented.
- PolarFire FPGA Imaging IP bundle: Image processing IPs for edge detection and alpha blending. Enables color, brightness and contrast adjustments. Includes the MIPI-CSI-2 camera interface.
These are all capable of being implemented on their FPGA video and imaging kit.
Development Aides
The PolarFire FPGA Video and Imaging Kit is an evaluation platform for Smart Embedded Vision designs. It enables evaluation of 4K image processing and rendering using dual camera sensors as well as numerous display interfaces.
PolarFire FPGA video and imaging kit. Image from Microsemi
The kit is designed to work with IPs mentioned earlier, as well as other popular video protocols.
The Partner Ecosystem
Microsemi maintains extensive network relationships across multiple disciplines. The goal of this alliance is to provide customers with trusted resources to expedite time to market.
The Microsemi Partner Network. Image from Microsemi
Members of the Microsemi Partner Network that figure prominently in Smart Embedded Vision Initiative include:
- Kaya Instruments is a new member of the Partner Network. It provides PolarFire FPGA IP Cores for CoaXPress v2.0 and 10 GigE vision. CoaXPress is a machine vision modality for medical systems and industrial inspection.
- ASIC Design Services Core Deep Learning (CDL) from ASIC Design Services is a scalable and flexible Convolutional Neural Network (CNN) solution for PolarFire FPGAs. It allows your trained machine learning inference engine to be run in PolarFire FPGAs.
As described by Shakeel Peera, vice president of product marketing for the FPGA business unit at Microchip’s Microsemi subsidiary, “Providing a suite of IP and hardware offerings alongside our partner ecosystem is essential to our clients’ ability to innovate while meeting their production schedules.” Further, “This is especially important because of the rapid evolution of machine and computer vision, driven by the growing adoption of AI, and the need to democratize edge-based vision systems.”
While there are many manufacturers who offer FPGAs, including Xilinx and Intel, Microsemi's new initiative represents a holistic approach to embedded vision. The goal seems to be to provide engineers with a one-stop shop for developing machine vision through FPGAs.
If you have experience using Microsemi's FPGA SoC tools, especially for vision applications, let us know about it in the comments below.
