In-house developed edge AI framework
you can implement on any device
Currently, existing AI frameworks are becoming segmented for different devices and platforms. With the inferencing-specialized AI framework ailia SDK, we offer a consistent and cross-platform inferencing environment. This is a world-leading AI development environment that realizes high-speed edge AI by connecting “AI models” that output various recognitions and results, and provides “execution platforms” for OSs and hardware.
Feature
Features & benefits of ailia SDK
01Delivers high-speed, cross-platform inferencing using CPUs and GPUs
In-house implemented accelerator with Vulkan, Metal, AVX and NEON enables high-speed inferencing that maximizes use of GPUs and CPUs. cuDNN can also be used if desired.
02State-of-the-art models are usable whether you’re on a PC or a mobile device
Behavior-tested open-source models for ailia SDK are now available. Supported by converted ONNX models, you can try the most advanced and varied models right away, regardless of device.
03Apps can be developed in various programming languages
Accommodates C++, Python, Java, as well as Unity (C#). You can develop apps for various industries, from communication to automotive and gaming.
04No need to install special drivers
As the model supports Vulkan (Windows/Android/Linux) and Metal (iOS, Mac), you don’t need to install special drivers and can effortlessly redistribute applications.
05Model compression and encryption makes it lighter and more secure
With proprietary weighting coefficient compression, you can save up to 1/3 in communication time and storage. Moreover, secure features are provided by encrypting the weighting coefficients.
Technology
Technologies in ailia SDK
SDK’s architecture
Explaining SDK’s API
Case study
Deployment examples