AI applications are beginning to be widely spread in our lives to such an extent that not a day goes by when we don't hear the word "AI".
Many people are interested in this technology, but when they actually try it out, they find out the tools and development flow are completely different from their current programming, which becomes a huge hurdle.
Renesas Electronics provides the RZ/V Series with an embedded AI processors, which can process real-time image with low power consumption same as conventional embedded MPUs.
We have released trained "AI Applications" that make it easier than ever to pre-trained AI using this RZ/V. In this blog, I would like to introduce these AI Apps.
AI Application development requires a lot of work, starting with preparation of an AI development environment, which includes selection of AI models, collection of training data, training of AI models, and so on. Furthermore, these tasks often have to be repeated many times to achieve a level of accuracy. When AI model can actually be used in user applications, the time and money spent here has been a major hurdle.
The newly released AI Apps can significantly shorten this AI development phase. In addition, since AI Apps are developed using open source and released on GitHub, which is free of charge, it can be used not only for evaluation purposes but also for your mass production development.
Let me introduce some of the AI Apps among over 20 AI Apps already available.
The first one is “Conference Room Usage Monitor”
It is a simple application that counts the number of head in the camera. By learning specifically for the human head, the number of people can be counted with a high accuracy. In addition to the application used for lateral shots, we will soon add "Elevator Passenger Counting" which allows head counting from an angle looking down from the ceiling.
The second one is “Defense wild animals for crop”
Crop damage amounts are more than 20 billion yen a year in Japan, which is caused by deer and wild boars (see Reference paper). As an ICT (Information and Communication Technology) tool for watching over large farms and ranches where human eyes are not enough, it would be great if we can reduce crop damages using this application.
We have also prepared a website for AI applications introduced here in an easy-to-understand manner. The customer only needs to get an RZ/V2L evaluation board. Why don't you try it out to develop image AI using RZ/V?
AI Applications and AI SDK for RZ/V
[Reference paper]
Responding to wildlife damage (Rural Development Bureau, Ministry of Agriculture, Forestry and Fisheries)
https://www.maff.go.jp/e/data/publish/attach/pdf/index-34.pdf