Defect Detection using Machine Learning

  Computer vision is a field of Artificial Intelligence, which gives the machines one of the five senses – the sense of sight. However, from a technical point of view it enables various machines to process visual inputs (digital images, videos, etc.), in order gather useful knowledge.

  Even though the computer vision field has many applications ranging from healthcare, where it is used in detection of anomalies in MRI and X-Ray scans, to transportation, where it is used in self-driving cars and pedestrian detection, the main subject of this article is defect detection.

  Defect detection is an application of the computer vision, which equips a system with different tools in order to detect abnormalities of objects: inconsistencies in dimensions, colours, shapes, parts which are bent, cracked, scratched or having misprints, etc.

  The main purpose of the defect detection application, in fields such as retail and manufacturing, is to create an automated system capable of aiding and even replacing, the human component in the process.

    

    The illustration below shows a few scenarios where the defect detection is needed:

  There are numerous benefits when an automated system for defect detection is created:

• Low Error Rate – since manual inspection requires the presence of a person and is subjective to the inspector, some imperfections can be attributed to human error.
• Faster – in comparison with the manual inspection, an automated system does not get tired and is not more likely to omit defects, as time goes on.
• Low Hazard Risk – inspections are executed in a lot of environments, which sometimes can be potentially hazardous to humans, however, for an automated system, the risk is non-existent.
• Precision – it is well-known that the human eye is incapable of making precise measurements, especially on a very small scale, therefore an automated system with multiple cameras will be more precise compared with a person.
• Cost of labour – an automated system, which is precise, fast, and has an low error rate is a costly acquisition, however in time, it costs less than the manual inspections.

  One of the creators of such automated system, is Mitutoyo, which equipped artificial intelligence with computer vision tools, to solve the problem defect detection with a high accuracy defect detection.

  On a final note, defect detection is one major point in the quality assurance of the products, and making it stricter by eliminating the elements prone to errors, can lead to the creation of almost perfect products.

References:

• Everything you need to know about Visual Inspection with AI, Jamshed Khan - https://nanonets.com/blog/ai-visual-inspection/
• Computer Vision - https://www.ibm.com/topics/computer-vision
• Defect Detection in Manufacturing With Unsupervised Learning, Viktoriia Akhremenko - https://mobidev.biz/blog/defect-detection-in-manufacturing-with-unsupervised-learning
• Sengottuvel, Ravikumar & K I, Ramachandran & Sugumaran, V.. (2011). Machine learning approach for automated visual inspection of machine components. Expert Syst. Appl.. 38. 3260-3266. 10.1016/j.eswa.2010.09.012.
• Mitutoyo AI Inspect - https://www.mitutoyo.com/aiinspect/