Image
Kazutaka Kikuchi
Kazutaka Kikuchi
Principal Business Development Specialist

RZ/V2MA introduction

AI-MPU RZ/V2MA is newly added on Renesas RZ/V series lineup.

https://www.renesas.com/rzv2ma

“A” for RZ/V2MA means accelerator and the AI MPU is designed for easy to add video analysis functions by endpoint AI on your current products via high-speed interface such as PCIe, Gigabit Ethernet or USB3.0. RZ/V2MA equipped DRP-AI, has been well received in high performance and low power, and newly added OpenCV accelerator to get more powerful vision processing capability.

1. Use case with USB camera 

Images from USB camera are analyzed by AI and failure detection processes by humans’ eyes are being automated in production line at factories. RZ/V2MA can treat maximum 4K image and realize high speed vision inspection machine w/ utilize DRP-AI and OpenCV vision processing. Low power DRP-AI makes the machine smaller and costly. Our partner Computermind Corp. prepared demo w/ RZ/V2MA.

English

https://youtu.be/V49JsFnomqI

2. Use case with Ether cameras

Video streaming from multiple IP cameras set on highway, parking lot or commercial facility are analyzed by AI and recognize number of number plate, counting cars passed the area or pose estimation of people in a facility are adopted rapidly. RZ/V2MA can handle multiple images via Ethernet with the built-in video codec, and AI analysis and image distribution can be performed for each image. Our partner amnimo Inc. has already released their product w/ RZ/V2MA and promotion videos here.

English

https://youtu.be/OUpY2w0VdNk

Japanese

https://youtu.be/E3Yc-_F2m9Q

3. Use case attached AI on your product via PCIe

Connecting the AI accelerator via PCIe is the easiest way to add AI function to your product that is already completed. RZ/V2MA supports PCIe endpoints and can be used as an AI accelerator.

For the RZ/V series, we have started offering "DRP-AI TVM (Note 1)" in addition to "DRP-AI Translator" as a tool to convert DRP-AI ONNX format AI models to RZ/V series. DRP-AI TVM is a tool based on the open source AI model compiler Apache TVM (Note 2), and it is now possible to accelerate a wider range of AI models with DRP-AI. Learn more about DRP-AI TVM here.
https://github.com/renesas-rz/rzv_drp-ai_tvm

(Note 1) DRP-AI TVM is powered by Edgecortix MERATM Compiler Framework
(Note 2) Apache TVM website: https://tvm.apache.org

Finally, we have also started selling the RZ/V2MA evaluation environment. If you are considering an AI product, please try the low-power AI inference performance of Renesas DRP-AI.

RZ/V2MA EVK

Image
RZ/V2MA EVK

Purchase link

http://www.shimafuji.co.jp/products/2688
 

Share this news on

Log in or register to post comments

Share this news on