- Himax, CEVA, emza Partner to Develop Low Power Vision Processing Platform
- Qualcomm's Latest Snapdragon Enhancements Leverage a Lithography Shrink
- New AMD Software Library, Hardware Support Deep Learning Acceleration
- Jeff Bier’s Impulse Response—Energy-efficient Processors are Critical for Embedded Vision
- AMD's ROCm: CUDA Gets Some Competition
Case Study: Deep Understanding of Processor Architectures and Computer Vision Algorithms is Key to a Breakthrough Product
Computer vision promises to be the key to the next set of "killer apps"—computer vision-enabled apps that will leverage artificial intelligence to help keep us safer and healthier. But computer vision algorithms are compute- and power-intensive, and need to process large amounts of data. These barriers have limited their use to enterprise, line-powered devices and cloud-assisted mobile devices. The implementation of computer vision algorithms on mobile processors, where compute resources are limited and power consumption must be curtailed, is key to wider use.
One important computer vision technology, 3D sensing, will spread to 80 percent of smartphones by 2018, earning a total of $2 billion by 2020, according to market researchers. The Lenovo Phab 2 Pro smartphone, which hit the market in late 2016, is the world's first smartphone to include Tango, a 3D sensing technology from Google, enabling a variety of games, utilities, and retail applications. Behind this launch was an intensive effort by BDTI to optimize the complex and demanding Tango algorithms to run efficiently and in real time on the Qualcomm Snapdragon 652 processor that powers the Lenovo phone.
BDTI's challenge in enabling real-time performance was to find ways to reduce the computational requirements of the algorithms without impacting the quality of results. BDTI studied the algorithms carefully and architected an implementation approach using image tiling and multi-threading. We then recoded the algorithms to implement this software architecture. Next, we optimized the algorithms to run efficiently on multiple processing engines in the Snapdragon 652, using hand-coded assembly, compiler intrinsics, refactoring, and other techniques. Thanks to the cumulative gains of all of these techniques, the optimized implementation achieved real-time performance—despite non-deterministic system behavior—as well as low power consumption.
BDTI's knowledge of embedded processor architectures, skills in architecting efficient software, and thorough understanding of computer vision algorithms were key to the successful implementation of Tango technology on the Lenovo Phab 2 Pro. To discuss how BDTI can help your company quickly transform your algorithms to meet tough processor performance and power consumption goals, contact Jeremy Giddings at +1 (925) 954 1411 or giddings@BDTI.com.