Seamless Integration Allows Prototyping of Finish-to-end AI Accelerated Imaginative and prescient Purposes with Assist for A number of CXP Cameras
WOBURN, MA, APRIL 9, 2024 — BitFlow introduced at the moment that it has efficiently built-in the unprecedented computing energy of the NVIDIA® Jetson AGX Orin module, which might carry out 275 trillion operations per second (TOPS), with the lightning-fast knowledge charges of its CoaXPress (CXP) body grabbers.
When mixed with the NVIDIA AGX Orin Developer Equipment, this cost-effective platform empowers engineers to prototype advanced machine imaginative and prescient and autonomous inspection purposes, leveraging AI accelerated picture processing whereas concurrently supporting as much as 4 CoaXPress (CXP) cameras and a number of concurrent AI software pipelines. Groundbreaking new purposes are extra simply developed that increase rule-based machine imaginative and prescient with image-based evaluation, making it doable to maneuver past “move/fail” to duties like picture classification, picture segmentation, and object detection.
As soon as proof-of-concept is established, a manufacturing mannequin can transfer ahead using an Advantech AIR-030 AI Inference System Field that includes PCI Specific x16 and primarily based on the NVIDIA Jetson AGX Orin. Because of this innovation, Time to Market and related improvement prices are considerably lowered.
50GB DATA TRANSFER BitFlow CXP body grabbers join on to the Jetson AGX Orin through a built-in x16 PCIe slot. Picture knowledge might then be transferred at speeds as much as 50GB per second from CXP cameras to the NVIDIA Ampere GPU structure — a lot quicker than what NVIDIA Jetson customers are sometimes restricted to utilizing USB3 or GigE Imaginative and prescient cameras. BitFlow CXP body grabbers DMA straight into the embedded GPU reminiscence for picture seize, pre-processing, and machine studying inference, shifting the load from the host laptop to keep away from CPU overhead.
Moreover quicker switch speeds, the CoaXPress interface permits a single cable to hold all knowledge, management, triggering, and as much as 13W of energy to linked cameras at lengths so far as 100 meters. CoaXPress eliminates the necessity for a number of cables and a neighborhood energy provide, subsequently giving the system integrator way more flexibility for his or her prototype designs.
Seamless integration between the NVIDIA Jetson AGX Orin and BitFlow body grabbers is achieved by BitFlow’s Linux AArch64 SDK. With the SDK being common, not solely is BitFlow’s full line of CXP body grabbers (Cyton and Claxon households) supported, however moreover the BitFlow Axion Digicam Hyperlink household is an possibility for patrons.
Study extra at www.bitflow.com.