Using Machine Vision to Extract Insights, Drive Efficiency

ADLINK, Touch Cloud, and Intel collaborate to bring artificial intelligence, image streaming, deep learning, and data analytics to the edge. The IoT solution can reduce production errors and streamline manufacturing processes, while smart cities can tap it to improve safety. 

 

Article Key

Image credit: Touch Cloud

Capturing video data can play an important role in IoT capabilities and services, but efficiently gathering, sorting, and classifying that data in real time requires highly advanced vision technologies. To that end, ADLINK, Touch Cloud, and Intel have developed an artificial intelligence engine to aid with data analytics, classification, and prediction for use in a range of industries, particularly in smart cities and manufacturing. The adaptable solution is integrated with AI and machine vision, which will bring inference acceleration to industrial and commercial technologies. 

Each company brings unique technologies to the solution. ADLINK supplies the optimized hardware platform and connectivity, and Touch Cloud provides the software application and analytics capabilities. The system is powered by the Intel® IoT Gateway processor, Intel® FPGA, and Intel®Movidius Myriad X Vision Processing Unit and is integrated with the Intel® Distribution of OpenVINO™ toolkit to enable smart vision application development. Together the technology allows companies to leverage their legacy infrastructure while implementing AI and gaining the benefits of IoT.

Actionable Intelligence

Edge computing allows manufacturers to turn vast amounts of machine- and device-generated data into practical, actionable intelligence at the source with minimal human intervention. In turn, this limits the amount of data that needs to be stored and transmitted to downstream systems, reducing the impact of network latency and cost.

Touch Cloud’s AI software engine handles the video data analysis and processing though the ADLINK edge computing platform. It supports near real-time video and image streaming with detection and trigger-event monitoring. The technology includes visual recognition for detection and classification and deep learning and AI model training for improved performance. Trained AI models can run on processors after being converted by Intel’s OpenVINO toolkit. The embedded FPGAs allow for industry-specific customization, making this technology adaptable for a wide range of applications.

Uses in Manufacturing and Smart Cities

AI-enabled machine vision can address scenarios that are too complex for rule-based machine vision. AI-enabled machine vision can mimic actions of the human eye and brain for faster processing of applications such as residential and commercial business meter reading, defect detection, and smart city uses. 

Automated Meter Reading. The process of manually checking meters at industrial jobsites is inefficient and costly. The MXE-210 inference platform and cameras from ADLINK, sitting at the edge, enable automated readings of meters connected to industrial gas, oil, and electricity lines. The collected data is sent in near-real-time to back-end servers via wired or wireless networks.

Image Classification and Segmentation. Using AI, images can be identified and separated into predefined categories. This capability allows for easier detection of product flaws and granular product segmentation and classification.

Automated Optical Inspection (AOI). To keep product yields and costs stable, defect detection needs to be accurate. AOI automates the visual inspection process during manufacturing by using cameras that constantly scan production lines for flawed products or improper skews. 

Defect Classification. Once defects have been identified, they can be classified to indicate type or origin. Companies can use this information to improve the quality of their products and implement changes to production methods if necessary. 

In addition, smart cities can use the technology with cameras to ensure essential services--including emergency responders, transit systems, and utilities--operate without fail. It can also be used to monitor and manage traffic flow or detect reckless driving. On a smaller scale, fleet managers can install cameras near truck or bus drivers and be notified if a driver engages in potentially hazardous behaviors. 

As IoT increasingly taps video to improve services, companies need vision products that will help organize, analyze, and index that data to bring operational improvements and increased profits. 

Learn More