Qualcomm Technologies and Baidu, Inc. announced they will work together to use the Qualcomm Artificial Intelligence (AI) Engine to drive conversion and application of Baidu PaddlePaddle open-source deep learning framework models on Qualcomm Snapdragon mobile platforms through the Open Neural Network Exchange (ONNX) interchange format. By utilizing both companies’ long-standing expertise in AI, this joint effort is intended to not only expand the overall AI ecosystem, but also help global developers and OEMs to more easily develop and launch AI features on devices powered by Snapdragon mobile platforms.
Both Qualcomm Technologies and Baidu are committed to optimizing neural networks on-device in order to drive the deployment of Baidu PaddlePaddle open-source deep learning framework to mobile devices powered by Snapdragon through the ONNX convertor.
“Qualcomm Technologies is excited to continue working with Baidu to advance on-device AI,” said Gary Brotman, director, product management, Qualcomm Technologies, Inc. “By combining the advantages of the Qualcomm AI Engine, ONNX and Baidu PaddlePaddle open-source deep learning framework, developers will have greater flexibility in creating and delivering AI-driven user experiences.”
“Through our cooperation with Qualcomm Technologies, we will invigorate the development of on-device AI applications. Baidu PaddlePaddle featuring ease-to-use, locality and rapid business integration will provide developers with higher-quality framework support and help developers with differentiated development,” said Wang Yi, head of PaddlePaddle open-source platform, Baidu.
Baidu PaddlePaddle is a deep learning platform independently developed by Baidu in 2013 to support various internal businesses. Baidu open-sourced PaddlePaddle in August 2016, which has shortly become an impressive open-source deep learning platform. The Baidu PaddlePaddle project can offer a deep learning framework and various additional technologies to simplify debugging and support large-scale training. Baidu also announced in February 2018 that it would fully support the Qualcomm AI Engine and its ecosystem.