Artificial intelligence, or AI, has seen a recent surge in use across a wide variety of industries. Anyone who has used Google or Amazon has been subject to one of the more popular uses of AI in the customized advertising. Another popular use of AI is image processing, which has seen an increased use in intelligence gathering and other security applications. Deep learning, a subset of AI, is particularly useful in image processing.
Deep learning algorithms use neural networking to process enormous data sets in order to carry out a desired task. For example, Japanese researchers used images of healthy tissue and cancerous tissue to teach a computer to detect early stage cancer using a deep learning algorithm. The computer detected cancerous tissue successfully about 80 percent of the time, which is comparable to a standard medical diagnosis.
Even beyond its security and medical applications, it is easy to see how deep learning could be applied to industrial vision systems. In very simple terms, a deep learning algorithm could automate quality control by learning what a “good” part looks like, and rejecting the rest. More broadly, deep learning algorithms could be used to learn what part of the manufacturing process causes a set of defects. The algorithm could then adjust the process and remove the defect from subsequent parts on the fly.
A simple example of process improvement is cookie baking. The deep learning algorithm could learn what a perfect cookie is, what a burned cookie is, and what an under-cooked cookie is. By actively monitoring the cookie batches using a vision system, the algorithm would be able to adjust the oven temperature as the cookies were moving through the process, saving time and money on bad batches.
The applications of deep learning are almost limitless with machine vision system. The ability of AI to process images is a well documented use of the technology, and most manufacturers have large databases of past material that could easily be used by deep learning algorithms for initial learning. Once the algorithm has learned, it can be maintained and adjusted over time just like any other piece of industrial equipment.
Learn more about Deep Learning Applications and Functions below.