Google Clips achieves intelligent image capture through built-in Intel Movidius VPU

Intel recently announced that Google’s latest Clips* automatic camera uses the Movidius™ Myriad™ 2 visual processing unit (VPU) to enable artificial intelligence processing at the camera terminal. Google Clips can automatically capture and manage dynamic photos of family, friends and pets, and these operations can be performed in the Clips application.

The Movidius Myriad 2 is a vision processing unit (VPU) that provides low-power, high-performance vision processing solutions across a variety of target applications, including embedded deep neural networks, pose estimation, 3D depth sensing, visual inertial odometry, and gesture/eye tracking.

With the built-in ultra-low power Movidius VPU, Google can run advanced machine learning algorithms in real time directly on the Clips camera. Thanks to onboard processing, Google Clips achieves lower energy consumption, shorter latency, and supports offline use. Juston Payne, product lead for Google Clips, said: "To realize the vision of Clips, we have been committed to integrating Intel's Movidius technology in Clips so that people can enjoy the many benefits of machine learning directly on their devices. Onboard processing can provide users with long battery life, enable them to quickly access their clips, and use the camera without an Internet connection. With Clips, parents and pet lovers can easily capture impromptu moments and record real-time life."

Movidius VPU technology provides a unique solution for low-power, high-performance computer vision intelligence to support the ever-evolving smart and autonomous devices. "As we worked with the Clips team, we saw how much intelligence Google packed into a device as small as Clips," said Remi El-Ouazzane, vice president of Intel's New Technology Group and general manager of Movidius. "This smart camera truly represents the level of onboard intelligence we dreamed of when we developed Myriad VPU technology."

“Our collaboration with Movidius means we can bring cutting-edge algorithms to production devices,” said Blaise Agüera y Arcas, director of machine intelligence products at Google. “With Clips, people can capture and share their moments in real time, anywhere, anytime, and Movidius technology greatly accelerates this vision.”

Last November, Intel completed the acquisition of Movidius, a leading computer vision company, and is committed to building a low-power, high-performance system-on-chip platform to accelerate computer vision applications and provide computer vision and deep learning solutions from devices to the cloud.

Movidius Myriad 2 visual processing units have long been applied to related industries. Spark, the smallest drone in DJI's product line, uses Intel's Movidius Myriad 2 visual processing unit, which optimizes the size and weight while having all the intelligent functions of other DJI drones. Zhejiang Dahua Company applies Movidius Myriad 2 visual processing units to video surveillance cameras. In addition to completing traditional tasks such as monitoring and recording, it also provides advanced video analysis functions such as crowd density monitoring, stereo vision, facial recognition, people counting, behavior analysis, and detection of illegally parked vehicles.