Intelligent Vision and Video

The ability to combine high-definition (HD) video, image processing, analytics, and connectivity has become an essential feature for creating next generation intelligent vision and video applications.

Computer vision software developers must use the appropriate development environment for the hardware architecture of choice to achieve the right combination of performance, cost, and power efficiency for the design’s use case. Use of artificial intelligence and deep learning technologies for video analytics have moved out of the laboratory and into commercial use for many end markets, such as security and surveillance, retail, and machine vision-based manufacturing.

Intel® Products enhances vision systems capabilities with heterogeneous camera-to-cloud inference and deep learning acceleration solutions using the following:

  • Silicon - Intel® FPGAs, IA CPUs, IA CPU with integrated graphics, and Intel® Movidius™ Vision Processing Units (VPUs)
  • Software and intellectual property (IP) - the Intel® OpenVINO™ toolkit which includes the Intel® FPGA Deep Learning Acceleration Suite

Intel® FPGAs and SoCs, along with IP cores, development platforms, and a software developer design flow, provide a rapid development path with the flexibility to adapt to evolving challenges and solutions in each part of the video or vision pipeline for a wide range of video and intelligent vision applications.

Real-Time Analytics

Intelligent, real-time video analytics are key functions in many vision applications. Typically, multiple algorithms such as motion detection, facial recognition, and object detection are needed to run simultaneously and across multiple camera sources. With Intel FPGAs and the Intel FPGA Deep Learning Acceleration Suite, designers can take advantage of industry-standard artificial intelligence frameworks, models, and topologies to create FPGA-powered versatile accelerators to implement convolutional neural network inferencing engines. By offloading the computation-intensive workloads from the Host CPU to the FPGA accelerator, designers can optimize the implementation of these multiple algorithms across the CPU + FPGA resources to improve overall system performance.

The FPGA Advantage at Each Step of the Video or Vision Pipeline

Intel® FPGAs offer performance, cost, flexibility, and integration advantages at each step of the video or vision pipeline. 

Flexible Sensor Interfaces

Image sensor suppliers often have proprietary interfaces, and the sensor interfaces keep evolving to keep up with improved sensor capabilities. Intel FPGAs enable designers to adopt new sensor interfaces easily without changing the rest of the design. In contrast, ASSPs require designers to adopt a new ASSP and redesign the camera board.

High-Performance Video and Image Signal Processing

Intel® FPGAs enable designers to design high-performance video and signal processing systems quickly and flexibly. IP cores such as Intel’s Video and Image Processing Suite allow designers to pick and choose video processing functionalities as needed, while the high-performance FPGA fabric can be leveraged to accelerate video pre-processing of high-resolution videos.

Universal Connectivity and Bridging

The abundance of different connectivity standards in today’s video and vision systems make FPGAs the ideal platform for universal connectivity and bridging.

Intel and our IP partners offer a variety of connectivity options, such as DisplayPort, high-definition multimedia interface (HDMI), and serial digital interface (SDI) IP cores, CoaXPress, Camera Link, GigE Vision, USB3.0, IEEE802.1x Time-Sensitive Networking (TSN) Ethernet, MIPI, and others.

 

 

High-Efficiency Video Compression

Highly-efficient compression standards such as H.264/H.265 allow the use of smaller FPGAs to reduce system cost.  Integration of video CODECs with other video processing functions on a single FPGA enables designers to achieve large bill of material (BOM) savings. Intel offers multiple video compression IP cores, which allow designers the flexibility to support multiple compression standards and profiles depending on the application requirements. Contact us to hear about one of the hottest trends in vision processing: cloud and network hosted digital media functionality.