CEVA has added support for the Open Neural Network Exchange (ONNX) format to the latest release of its CEVA Deep Neural Network (CDNN) compiler.
ONNX is an open format created by Facebook, Microsoft and AWS to enable interoperability and portability within the AI community, allowing developers to use the right combinations of tools for their project, without being ‘locked in’ to any one framework or ecosystem. The ONNX standard ensures interoperability between different deep learning frameworks, giving developers freedom to train their neural networks using any machine learning framework and then deploy it using another AI framework. The ONNX support allows developers to import models generated using any ONNX-compatible framework, and deploy them on the CEVA-XM vision DSPs and NeuPro AI processors.
"CEVA is fully committed to ensuring an open, interoperable AI ecosystem, where AI application developers can take advantage of the features and ease-of-use of the various deep learning frameworks most suitable to their specific use case,” said Ilan Yona, vice president and general manager of CEVA's Vision Business Unit. “By adding ONNX support to our CDNN compiler technology, we provide our CEVA-XM and NeuPro customers and ecosystem partners with much broader capabilities to train and enrich their neural network-based applications.”
The CEVA Deep Neural Network (CDNN) creates fully-optimized runtime software for the DSPs and NeuPro AI processors. Targeted for mass-market embedded devices, CDNN incorporates a broad range of network optimizations, advanced quantization algorithms, data flow management and fully-optimized compute CNN and RNN libraries into a holistic solution that enables cloud-trained AI models to be deployed on edge devices for inference processing.