How to identify opportunities for AI in machine vision?

Artificial intelligence (AI) is being adopted by industries to harness the power of data and use it to make smarter decisions.

This article will explain how to identify opportunities for AI in machine vision applications.

Business Requirements for Artificial Intelligence Systems

Managing Expectations

AI approaches have specific use cases. After all, it is not a universal solution and cannot solve all problems. Some applications are better suited to traditional computer vision, some may require both, and some may require only artificial intelligence. AI systems are expensive—both in terms of cost and upfront resources required. Open source tools require significant development time, and external tools are often expensive. Additionally, a GPU is often required to achieve adequate performance on the system. Many manufacturers often don't have GPUs or equivalent processing power. Therefore, it is important to determine which applications are well suited for AI with strong business needs.

The importance of visual system settings

Before entering AI, it is recommended to have a solid foundation in visual system settings. However, this is less important for AI, which can often handle worse conditions than traditional systems. All the normal machine vision system rules apply here - good lighting, camera resolution, focal length, etc. If any of these factors aren't up to scratch, it's worth going back and addressing them before delving further into AI. Ensure robust vision system setup for best results.

Reference Human Performance

AI systems are most successful where human performance is strong. Once the system is set up, operators can easily identify/classify images by eye, thus determining whether they are suitable for AI. However, if human performance is insufficient, then the AI ​​model is likely to perform poorly. Using human performance as a reference point for what an AI model can achieve, if an operator can only identify images 70% of the time correctly, it's unlikely that the AI ​​will perform better than that. Therefore, if human performance is not good enough for an application, that performance issue should be addressed first and improved to an acceptable level. Once operators achieve expected performance, AI can be considered.

Time and Resources

Collecting images and training the model requires considerable effort. Often, collecting high-quality images is the hardest part because many manufacturers have very low defect levels. Without data, it can be difficult to train a model for defective parts. Training tools are helpful, providing pre-trained models that require fewer samples to train. Training is an iterative process spanning multiple steps to find the ideal parameters for the model to run. Optimizing a model often requires time and experimentation. Additionally, if new data appears in the field, the model will need to be trained and deployed again.

Artificial Intelligence Application Examples:

One example application of artificial intelligence in machine vision is for final assembly inspection, another is for printed circuit boards or PCB detection.

❶ Final assembly inspection:

Background

Final inspection of parts/products or components is usually performed by operators, or traditional machine vision system, or both. Teledyne cameras will be highlighted here as an example product. The final inspection might check for bent pins, scratches on the surface, proper placement of connectors, alignment of stickers, proper printing of text, distance between mechanisms, and more. Basically, any exceptions that occur during the build process need to be found. But then the list of criteria that needs to be looked up quickly becomes very long. Traditional rule-based systems struggle to handle all corner cases, and training new operators is difficult.

Why AI?

There are often too many rules to determine what a "pass" is. This makes it difficult for traditional machine vision systems to achieve good performance. The alternative is that manual inspection is time-consuming for many companies and difficult for new operators to make some ambiguous judgments. Traditional rule-based systems often do not have adequate performance, and manufacturers rely on operator judgment to help. There may be different lighting conditions, as well as high variations in defect location, shape, and texture. Often, a simple "good/bad" qualitative output is all that's needed. However, this can also be combined with traditional rule-based algorithms if desired.

benefit

With AI, setup is much easier. After collecting a large number of images to train a model, getting a system running usually requires far less development work than a rules-based system, especially using AI tools. With a suitable system, usually using a GPU, the check is much faster, on the order of milliseconds. If provided with good data, AI systems should also perform more reliably than humans and are a good way to standardize inspection procedures. The algorithm is typically trained on data provided by multiple operators, which can reduce human error. This helps mitigate human bias or fatigue that may arise from a single operator. In this example, AI can help manufacturers reduce out-of-box failures and improve inspection quality and throughput.

❷ PCB Inspection:

Background

PCB manufacturers need to inspect their circuit boards for any defects. It may be a bad solder joint, short circuit or other abnormality. AOI (Automated Optical Inspection) machines are usually used. However, since defects vary so much, it is difficult to handle all edge cases. And the performance of rule-based systems is not accurate enough, and manufacturers will ask operators to perform manual inspections, which is time-consuming and expensive.

Why AI?

It is difficult for traditional AOI systems to identify defects. It either overshoots or underperforms, causing a defective PCB to pass or a good PCB to fail. Similar to other situations, there are too many rules to determine a "good board". Depending on the application, AI can be used here to classify defects that vary widely in size and shape, such as short circuits, opens, faulty components, welding defects, etc.

Benefits

With artificial intelligence, manufacturers can improve the accuracy and quality of inspections. This helps reduce the number of defective PCBs passing inspection. It also saves the time and labor costs of any manually assisted inspections and increases throughput by automating tasks that take operators longer to complete.

The above is the detailed content of How to identify opportunities for AI in machine vision?. For more information, please follow other related articles on the PHP Chinese website!