Qualcomm unveiled a deep learning software development kit (SDK) which makes it easier for developers to bring complex machine learning tasks to Snapdragon 820 devices such as smart phones, security cameras, automobiles and drones, all without a connection to the cloud. This kit can implement deep learning tasks such as scene detection, text recognition, object tracking and avoidance, gesturing, face recognition and natural language processing, said Qualcomm.
“The SDK can target anything that requires an embedded Snapdragon 820 chip, such as healthcare, automotive, security and imaging. The kit will allow companies to run their own proprietary trained neural network models on portable devices”, says Qualcomm.
The SDK, called the Qualcomm Snapdragon Neural Processing Engine, is powered by the Qualcomm Zeroth Machine Intelligence Platform, a software platform designed for mobile machine learning.
Zeroth is currently being used for two technologies: Snapdragon Scene Detect, which classifies objects, items, and people within a visual scene, and advanced malware detection software found in Snapdragon Smart Protect, which uses machine learning to look for suspicious behavior.
“The demand for untethered, mobile learning-driven user experiences is increasing rapidly and Qualcomm Technologies’ customers have been clamoring for tools to help them realize their product visions. With the introduction of the new Snapdragon Neural Processing Engine SDK, we are making it possible for myriad sectors, including mobile, IoT and automotive to harnesses the power of Qualcomm Snapdragon 820 and make high-performance, power efficient on-device deep learning a reality”, said Gary Brotman, director of product management, Qualcomm Technologies.
As mentioned in the announcement, the SDK offers accelerated runtime for on-device execution of convolutional and recurrent neural networks on the Snapdragon 820 cores (Qualcomm Kryo CPU, Qualcomm Adreno GPU, Qualcomm Hexagon DSP), and support for common deep learning model frameworks, including Caffe and CudaConvNet.
Comments Frederick Soo, chief technology officer of NAUTO, the autonomous vehicle technology company: “The Neural Processing Engine SDK means we can quickly deploy our proprietary deep learning algorithms to our Snapdragon-based connected camera devices in the field, which can detect driver distraction and help prevent auto accidents. That same deep learning helps NAUTO’s system deliver valuable information that helps our customers reduce liability and drive more safely and efficiently.”
The Snapdragon Neural Processing Engine SDK is expected to be available for Snapdragon 820 processors in the second half of 2016.
For further details, view the full Press release.