Machine Vision Technology in Production: Use Cases

We at TaQadam produce different computer vision technologies. In this blog we tell about using machine vision in production for some common use-cases.

What Machine Vision is and where it is used

Machine vision is a technology in the field of artificial intelligence for image recognition. In general, the tasks of machine vision systems include the capture of the images on the device, processing on the edge, and link it to machine response.

Computer or machine vision technologies are used across different industries: from traffic control and public services, medical imagery, and robotic processing at industrial sites.

As Industrial 4.0 revolution penetrates across our lives, there is an increasing interest in machine vision AI, IOT (internet of things) for optimizing the manufacturing production. Today, machine vision systems are an effective tool for controlling performance, detecting anomalies, and reducing human supervision at site.

Use cases: automating production

Sensors and industrial cameras have opened up new perspectives in the automation of the production process:

• measurement (measuring/testing specific product dimensions)
• product verification (checking the correct position of all parts on the assembly)
• product identification (barcode and 2D code reading, product type determination based on shape, size, and other characteristics)
• product sorting (for example, sorting packages by color, etc.)
• navigation system (determining the position of the product on the assembly line and using its coordinates for IOT device).

Use case from: Blumenbecker group

Use cases: safety and protection​

In the field of industrial safety, there are several important tasks.

Perimeter control as a measure against crime and unlawful trespassing. These video analytics has been around for years. But with computer vision AI, there is a likelier chances it to be detected early.

Safety movement control. In the companies of chemical or metal production, the departments of Hazardous Materials Safety, has been implementing the monitoring systems using AI. With the better control, there is a reduced risk of employees to avoid injuries, in loading or uploading or approaching certain areas. In some advanced cases, sites are using historical video analytics to design the paths for optimized and safe performance at production.

Another example of a safety-related application is detecting helmets or safety gear. Computer vision and machine learning have made it possible to automatically identify non-compliance.

All these techniques lead us to the case, where AI will also bring a significant data value to insurance companies. What is more, it can become easier to solve insurance cases.

Use cases: quality control

Machine vision in production can also be used for quality control. This task can be divided into two components.

One is related to physical quality control. You can view the dimensions of certain objects. Most often, this concerns small things: some caps from milk bags or bottles. They have a fairly simple cheap production process, a lot of defects, they just need to be filtered out, it is not profitable to make details better.

The second task is to determine whether the actions are performed correctly. Using the video camera, you can automatically see what events are happening, and track the correctness of the employee’s actions, as well as monitor the speed of work.

Use cases: anomaly detection

Machines are replacing humans often when it comes to repetitive work.

An engineer who spends hours at work analyzing the defective parts, for example, is prone to make more mistakes. This is the situation in which machine vision is essential to increase the overall performance. Computer vision based on machine learning can automatically detect defects and anomalies at the infinitely larger number.

Sensors scan all the details and send a signal if something is wrong. Cameras paired with LEDs carefully study the image and transmit the images to the computer. It already has a large database trained on anomalies recognized in the past, and it will instantly give the command to the operator further down the pipeline to sort them out.

Our work at industrial Machine Vision ​

Helmets detection ​

Details detection

Forklifts detection

Machine vision allows to reduce the cost of production (the system performs the operations that people used to do, works without interruption) and simultaneously improve the quality of the operations performed, which, in turn, improves the quality of the product itself. Using machine vision allows to track the quality of items that have a complex shape, large or very small area. In general, investing in the creation of a machine vision system should be considered as the first step towards high-tech production.

Originally published at: https://taqadam.io/machine-vision-in-productio