Arm on Tuesday introduced Project Trillium, a suite of machine learning IP that aims to power neural engines as they migrate to the edge, and AI applications that demand local processing.
The IP includes the company's machine learning and object detection processors and neural network libraries.
Arm says its machine learning processor was built with a ground-up design enabled by open-source software for high performance efficiency and inference at the edge. The processor analyzes fewer pixels for faster, fine-grain object recognition, and delivers performance of more than 4.6 TOPs with efficiency of 3 TOPs per watt.
The object detection processor was designed to objects in images and rich characterizations in real time with full HD at 60 frames per second. Arm said the first generation OD processor powers the latest Hive security camera, but the company listed a bevy of future uses cases for the combined solution.
In addition to smart cameras and other vision-based devices, Arm said the ML and OD processors can be applied to smart city systems for real time information and control, including pedestrian impedances, congestion, and safety issues. The company also said the underlying architecture will scale all the way to the data center.
"The level of sophistication of edge device computing moved faster than anyone expected," said Rene Haas, president of Arm's IP Products Group. "There will be a lot of different applications as machine learning moves the edge."
As for the neural network libraries, Arm said Trillium provides a seamless link into a bank of Arm partners delivering neural network apps, including frameworks such as Google TensorFlow, Caffe, AndroidNN API and MXNet.
"We are launching Project Trillium two weeks before Mobile World Congress 2018 kicks off," said Jem Davies, GM of machine learning at Arm. "This year's event halls will feature early implementations in the form of the Arm Object Detection processor in products such as IP security and smart cameras. Future shows will see the fullest range of Arm ML technologies gaining traction far wider and supporting the growth of smart connected devices as part of a new world built on AI."