Artificial intelligence extends the capabilities of machine vision

Industrial vision systems using image processing algorithms have been deployed for many years for automated inspection operations on production lines to detect anomalies, contaminants, functional defects and other irregularities on production lines. manufactured products. Cognex details how image processing tools based on artificial intelligence technologies are now providing them with new weapons to improve performance and extend their spectrum of use.

Artificial intelligence includes any technique that aims to reproduce the behavior of living beings by machines, without this behavior being explicitly programmed. Learning takes place empirically, by feeding the machine with a multitude of examples, using algorithms that allow, through calculations on data, to take or help to make decisions. Artificial intelligence is the transposition to the machine of the notion of reasoning and choice in autonomy. It therefore enables computer systems to learn from data and examples in order to predict outcomes.

According to Forrester Research, 53% of tech decision makers are implementing or expanding their use of artificial intelligence, and 20% plan to implement artificial intelligence in the next 12 months. Investment in artificial intelligence solutions deployed in manufacturing is expected to increase by almost 50% each year to reach $ 17.2 billion by 2025.

Artificial intelligence can be exploited in areas as diverse as supply chain management, quality testing and inspection, and predictive equipment maintenance. It is therefore called upon to deploy in industry and to transform the management of production processes. However, at a time when this technology is progressing rapidly and becoming more user-friendly, many companies still find it difficult to take full advantage of it, in particular due to its cost, the time to set up, the necessary expertise and the reliability of the results.

By redefining their performance expectations, whether it's detecting faults, avoiding false rejects or saving time, manufacturers who embrace artificial intelligence, and especially deep-based applications learning, in the context of automated inspection applications, should nevertheless benefit from this. A deep learning-based project can save money, improve efficiency, and gain insight into their own production process. If there are any direct upfront costs associated with implementing a deep learning-based solution, including software and hardware expenses, development and engineering costs, and the time required to collect data from Entrance.

Cost reduction

Manufacturers who plan to replace manual inspection operations put in place where traditional industrial vision was too difficult to implement, will see their costs reduced in return. The cost of manual inspection is dominated by the expenditure on labor. However, human visual inspection prevails in situations which require learning from examples and determining acceptable deviations from the control system. Inspectors are good at distinguishing cosmetic from functional defects, as well as determining variations in appearance of parts that can affect perceived quality. Although their working speed is limited, inspectors are able to conceptualize and generalize. They excel at learning from examples and are able to determine acceptable slight discrepancies between parts. This makes human vision the best choice, in many cases, for the qualitative interpretation of a complex and unstructured scene, especially those with subtle and unpredictable flaws.
Human inspectors are therefore more efficient in certain situations than automated control solutions, provided they show total attention. However, most operators can hardly concentrate and maintain the necessary attention for a long period of time. In addition, the results of the checks vary from one individual to another. This leads to inconsistencies when changing shifts or between different production lines. In contrast, a machine vision solution offers the speed and reliability that only a computer system can. It excels in the quantitative measurement of a structured scene due to its speed, precision and repeatability. A system with machine vision lenses and cameras can inspect details of objects invisible to the naked eye, with superior reliability and a reduced error rate. On a production line, machine vision systems can check hundreds of parts per minute reliably and repeatedly, far exceeding the inspection capabilities of operators. When they implement deep learning tools, then they offer the flexibility of human visual inspection with the speed and reliability of a computer system. Deep learning technology uses neural networks that mimic human intelligence to distinguish anomalies, parts and characters, while tolerating natural variations in complex patterns. with higher reliability and lower error rate. On a production line, machine vision systems can check hundreds of parts per minute reliably and repeatedly, far exceeding the inspection capabilities of operators. When they implement deep learning tools, then they offer the flexibility of human visual inspection with the speed and reliability of a computer system. Deep learning technology uses neural networks that mimic human intelligence to distinguish anomalies, parts and characters, while tolerating natural variations in complex patterns. with higher reliability and lower error rate. On a production line, machine vision systems can check hundreds of parts per minute reliably and repeatedly, far exceeding the inspection capabilities of operators. When they implement deep learning tools, then they offer the flexibility of human visual inspection with the speed and reliability of a computer system. Deep learning technology uses neural networks that mimic human intelligence to distinguish anomalies, parts and characters, while tolerating natural variations in complex patterns. Machine vision systems can check hundreds of parts per minute reliably and repeatedly, far exceeding the inspection capabilities of operators. When they implement deep learning tools, then they offer the flexibility of human visual inspection with the speed and reliability of a computer system. Deep learning technology uses neural networks that mimic human intelligence to distinguish anomalies, parts and characters, while tolerating natural variations in complex patterns. Machine vision systems can check hundreds of parts per minute reliably and repeatedly, far exceeding the inspection capabilities of operators. When they implement deep learning tools, then they offer the flexibility of human visual inspection with the speed and reliability of a computer system. Deep learning technology uses neural networks that mimic human intelligence to distinguish anomalies, parts and characters, while tolerating natural variations in complex patterns.

Join