Embedded AI applications are becoming more and more popular. It needs a little imagination to become a reality.
Nowadays, many of the devices we use every day make use of neural network technology: they support photo search in smartphones, and face recognition helps people use their faces to unlock their phones; they can know how to drive by identifying license plates. Enter or leave the congestion zone. They are "smart" enough to detect a person's body language - which is very useful for surveillance systems, and they are also applied to AV systems with cameras that recognize each person and give the system personalized content services. Applications involved in automobiles include lane departure warning, driver status monitoring, etc. In addition, a large number of experiments and applications have proved that neural networks are better at detecting skin cancer than humans - the application of visible neural networks is both extensive and diverse.
Different from “traditional” cloud-based artificial intelligence (AI), these neural network computing based on field application devices can be called embedded AI; this is a market that is rapidly developing but also eager for technological innovation, and also many SoCs. Design companies are pioneering markets with innovative products. With Imagination's launch of the Series2NX Neural Network Accelerator (NNA) in September last year, the rules of the game in the field of embedded AI acceleration are being changed. Now, let's take the two core products released by Imagination as an example to analyze the comprehensive performance indicators of embedded AI in different application markets. The two products are the PowerVR AX2185 for the high-end market and the AX2145 for the mid-range market. Below we will introduce to you in detail.
As a tool to solve problems and drive new applications in multiple fields, neural networks are becoming more and more popular. The following figures illustrate the rapid development of embedded AI applications in the fields of mobile devices and intelligent surveillance cameras.
In the face of the rapidly evolving embedded AI market, with the launch of new engines such as Imagination's two new core products, the industry will be able to implement low-cost AI functions on end devices to provide consumers with more new AI applications. program. These cores will be able to meet the performance requirements developers need to create applications, truly realize the changes brought by artificial intelligence, and these improvements will soon become a reality, it will involve multiple application markets such as smartphones, smart Cameras and cars, etc. The possibilities of the future are infinite, and the only thing that needs is a little imagination (Imagination).