Ever sit at a red light and wonder how the light system knows that you are there waiting to go? Or wonder how your cell phone can pick up on objects around the house and search for them on the internet? This is called Computer vision, and it changed how we search and shop for items, manufacture parts, and travel around town. It’s an Artificial Intelligence (AI) system designed to allow computers to observe and understand the surroundings to make the best judgment on what is being asked. I will be talking about what it is, how it works, and the applications we use it on.
Computer vision, Artificial Intelligence (AI), permits computers and systems to get important information from digital images, videos, and other visual inputs to require actions or make recommendations based on that information. AI enables computers to think, and computer vision allows them to see, observe and understand. Computer vision works just like human vision by scanning and assessing an object. Our eyesight has decades’ worth of knowledge to back them up when it comes to deciding on what the object we are looking at is, how far we are from it, and if it is moving or not.
This technology executes these actions faster with cameras, data, and algorithms. This technology allows us to inspect and analyze thousands of products or processes a minute, allowing us to go beyond human capabilities by quickly noticing defects or issues. Industries are using this technology for maintaining energy and utilities. Such as IT solutions for wind energy producers that regular drone images are used to identify maintenance needs to include but are not limited to dirt accumulation on wind turbines, dead insects, salf (offshore wind turbines), and cracks in foundations or towners. It is also being used in manufacturing and automotive to act like a complete sensory apparatus that simultaneously takes in the environment around you and analyzes it for potential threats, obstacles, and other relevant situations needed to react to while driving a vehicle.
Computer vision has been in the works for about 60 years. It began around 1959 when neurophysiologists showed a cat an array of images, attempting to link a response in its brain. They found that the cat responded first to hard edges or lines, which meant the image process starts with simple shapes. At the same time this occurred, the first computer image scanning technology was developed, allowing computers to digitize and procure images.
Around 1963 another milestone was reached when computers could transform two-dimensional images into three-dimensional ones. Marking the beginning of the AI quest to solve the human vision problem, AI emerged as an academic field of study. Then in 1974, the introduction of optical character recognition (OCR) technology and intelligent character recognition (ICR) was discovered. OCR technology could recognize text printed in any font or type. ICR could decipher hand-written text using neural networks. Both found their way into documents and invoice processing, license plate recognition, mobile payments, and other typical applications.
Let’s move ahead to the year 2000 when the study of object recognition started. 2001, the primary real-time face recognition applications appeared. 2010, the ImageNet data set went live, which contained many tagged images across thousand object classes and provides a foundation for convolutional neural network (CNN) and deep learning models used today. Lastly, in 2012 a team from the University of Toronto entered a CNN into an image recognition contest. The model was called AlexNet, which drastically reduced the error rate for image recognition to just a few percent.
Now you and I both know that no one likes cliffhangers in shows, especially a really good one. Well, stay tuned for part two in the talk about computer vision coming 12/23/21.