Cadence and ArcSoft collaborate on AI vision applications including facial unlock

Jul 24, 2018, 9:00 am EDT | Stephen Mayhew

Electronic design company Cadence Design Systems and imaging intelligence company ArcSoft have developed a suite of AI and vision applications for Cadence Tensilica Vision DSPs (digital signal processors), and the joint solution is now shipping in smartphones, having been built into an applications processor from a leading global provider, according to the announcement.

Beauty shot, high dynamic range (HDR), bokeh, and facial unlock applications have been ported to the Vision P6 DSP. ArcSoft says its facial recognition algorithms are built for high efficiency, low power consumption and high accuracy, meeting facial unlocking requirements for fast response time at low power. The other capabilities likewise benefit from lower power consumption while improving performance with AI.

“We needed a very high-performance vision and AI DSP to accelerate our beauty shot, HDR, bokeh and facial unlock applications,” said Frison Xu, marketing VP at ArcSoft. “Cadence DSPs come with excellent software tools and a comprehensive vision and AI library which allowed us to increase the performance of our HDR algorithm by 6X compared to CPUs. Cadence’s low-power, high-performance Vision P6 DSP is an ideal DSP for porting our vision and AI applications.”

Two of the top five mobile application processor vendors have designed the Vision P6 DSP into their products, allowing them to take advantage of its flexible hardware options and an extensive library of vision and imaging DSP functions and applications from its ecosystem partners. The Vision P6 DSP also supports AI applications developed in the Caffe, TensorFlow and TensorFlowLite frameworks through the Tensilica Xtensa Neural Network Compiler, and supports the Android Neural Network API for on-device AI acceleration.

“The combination of Cadence’s vision and AI DSP and ArcSoft’s software technology make on-device vision and AI experiences possible on today’s mobile platforms,” said Pulin Desai, product marketing director, Tensilica Vision DSP product line at Cadence.

The Korea Advanced Institute of Science and Technology (KAST) developed an ultra-low power consumption AI chip for facial recognition last year, but they were not designed to be built into smartphones.