Cognex presents an intelligent industrial camera optimized by deep learning software. This combination multiplies the possibilities for efficient and accurate inspection in industrial automation.
Deep learning and artificial intelligence (AI) are buzzwords in industry and many other fields. The principle behind deep learning, a type of AI that exploits neural networks, is quite simple: just as a human being learns what a house is from a large number of examples, a deep learning system eventually identifies the nature of an object from images. Whether it's the bumper of a car, the envelope of a package or something else, deep learning offers an example-based method of solving inspection problems. What's more, each inspection can be continuously improved by feeding the algorithm with more example images. This new data improves the system, leading to more accurate results.
Pushing back the limits of automated inspection
Manufacturers continue to rely on human inspectors throughout the production process, as conventional machine vision systems are unable to handle the complexity or variability of certain tasks. For their part, human beings produce variable results from person to person, and can become fatigued as the day progresses. What's more, for a variety of reasons, many aspects of production remain uninspected. Both of these realities expose manufacturers to quality control problems or recall risks.
A vision system with integrated artificial intelligence is capable of combining the judgment capabilities of a human inspector with the robustness and consistency of an automated inspection or quality control solution. According to Cognex, this is precisely the main advantage of the new Cognex In-Sight D900 vision system.
The first of its kind, according to the manufacturer, this solution integrates In-Sight ViDi, Cognex's deep learning software, into an industrial-grade smart camera. This combination multiplies the possibilities of in-line industrial inspection.
Equipment: unrivalled versatility
The In-Sight D900 hardware is a compact, industrial-grade smart camera that can be easily installed and deployed on the line without the need for a PC. Highly modular, the IP67 vision system includes interchangeable lighting, lenses, filters and covers that can be customized to meet the needs of each application.
The High Dynamic Range (HDR+) imaging system creates images with homogeneous exposure, an LED indicator enables remote control of inspection success or failure, and results can be stored locally on an SD card. The integrated inference engine has been added to specifically run deep learning applications. While the hardware is among Cognex's most advanced, it's the In-Sight D900's ability to run a new class of deep learning-based applications that makes it stand out.
Deep learning and easy application development
The In-Sight D900 is based on a familiar, user-friendly spreadsheet interface, simplifying application development and factory integration. Technical sales engineers have access to a full range of conventional machine vision tools such as PatMax, edge probes and measurement tools. However, with In-Sight ViDi, the D900 features three deep learning tools, all targeting specific and widespread applications: ViDi Read, ViDi Check and ViDi Detect.
These new inspection tools help industrial automation customers easily solve applications that are too time-consuming or complex to be deployed with conventional rule-based machine vision tools.
For all three application tools, users can take advantage of In-Sight's intuitive spreadsheet interface, which enables rapid configuration of deep learning applications without programming. The In-Sight spreadsheet simplifies application development and factory integration with a comprehensive set of I/O and communication functions. It also makes it possible to combine classic Cognex rule-based vision tools (such as PatMax Redline) and deep learning tools within the same task, speeding up deployments.
Because In-Sight ViDi requires much smaller image sets and shorter learning and validation periods than other deep learning solutions, applications are quick and easy to set up, train and deploy.
The In-Sight ViDi Read tool solves complex OCR applications in a matter of minutes. This module deciphers severely distorted and badly engraved codes using optical character recognition (OCR).
Ready to use, this tool reduces development times, thanks to a font library pre-trained via deep learning. All the user has to do is define the desired region and font size. If new characters are introduced, this robust tool can be trained to read application-specific characters that cannot be decoded by conventional OCR tools. Compared to conventional vision systems without deep learning, this feature offers clear advantages.
In-Sight D900's ViDi Check tool enables manufacturers to check assemblies quickly and accurately. The system is able to detect complex features and objects. It checks whether parts and kits are correctly assembled according to their location within a user-defined layout. The tool can be trained to create an extensive library of components, which can be located in the image even if they appear at different angles or vary in size.
A third tool, ViDi Detect, has been designed to analyze complex defect detection tasks. It is capable of learning from images of compliant parts to identify defective ones. In-Sight ViDi Detect is ideal for spotting anomalies on complex parts and surfaces, even in situations where defects may be unpredictable in appearance.
A new deal for visual inspections
The integration of In-Sight ViDi's deep learning-based vision software with a smart camera helps industrial automation customers solve complex OCR, assembly verification and defect detection applications, which are often too difficult to deploy with conventional rule-based machine vision tools and require reliable, fast and consistent results impossible to achieve with human inspection.
With the introduction of this system, Cognex covers three application areas: optical character recognition (OCR), assembly verification and defect detection. While others are likely to follow, these three areas open up a wide range of possible tasks that the new system can handle efficiently, with greater precision than a conventional vision system. In terms of markets, the In-Sight D900 can be used in a number of sectors, including automotive, consumer electronics, FMCG, packaging, food, medical equipment and logistics.