Intels Brings Out AI Camera, Xeon Scalable Optimization And FPGA-based Accelerations For Deep Learning At Baidu Create

Recently, at the Baidu Create held in Beijing, the vice president of Intel announced a series of collaboration with Baidu on AI (Artificial Intelligence). The series included Intel power Baidu’s Xeye which is an advanced AI retail camera with Intel Movidius vision Processing Unit that highlighted the plants of companies to accelerate workload. This VPUs will be offered as a service utilising Intel FPGAs, it also optimises PaddlePaddle* one of the deep learning Framework from Baidu for intel Xeon Scalable Processors.

Fig. 1: Baidu AI

(Image source: Intel Newsroom)

As Gadi Singer, the vice president and architecture General Manager of Intel’s Artificial Intelligence products, puts it, “From enabling in-device intelligence, to providing data centre scale on Intel Xeon Scalable processes, to accelerating workloads with Intel FPGAs, to making it simpler for PaddlePaddle developers to code a cross platforms, Baidu is taking advantage of Intel’s products and expertise to bring its latest AI advancements to life.”

The Baidu’s Xeye camera makes use of Intel Movidius Myriad to VPUs For delivering high performance and low power visual intelligence for retailers. The credit for this goes to Intel’s purpose-built VPU solutions teamwork Baidu’s highly developed machine learning algorithms, this camera is capable of analysing gestures and objects while detecting individuals to offer personalised shopping experiences in retail settings.

Baidu Is currently developing a completely heterogeneous computing platform founded on latest Field Programmable Gate Array (FPGAs) technology by Intel. Intel FPGAs can accelerate performance as well as Energy Efficiency while adding extra flexibility two data centre workload. It would also allow workload acceleration as a service over Baidu Cloud.

With PaddlePaddle that has now been optimised for intel Xeon Scalable processors, data scientist as well as developers can now make use of the same hardware that empowers data centres and clouds across the world who developed the AI algorithms. PaddlePaddle has been optimised for intel Technologies at different levels including communication, architecture, memory, and computing.

The two legendary companies are also looking into integration of nGraph and PaddlePaddle, a Framework neutral, DNN (Deep Neural Network) model compiler is capable of targeting a wide range of devices.