Computer Vision
What it is and why it matters
Computer vision is a field of artificial intelligence that trains computers to interpret and understand the visual world. Using digital images from cameras and videos and deep learning models, machines can accurately identify and classify objects — and then react to what they “see.”
History of computer vision
Early experiments in computer vision took place in the 1950s, using some of the first neural networks to detect the edges of an object and to sort simple objects into categories like circles and squares. In the 1970s, the first commercial use of computer vision interpreted typed or handwritten text using optical character recognition. This advancement was used to interpret written text for the blind.
As the internet matured in the 1990s, making large sets of images available online for analysis, facial recognition programs flourished. These growing data sets helped make it possible for machines to identify specific people in photos and videos.
Today, a number of factors have converged to bring about a renaissance in computer vision:
Mobile technology with built-in cameras has saturated the world with photos and videos.
Computing power has become more affordable and easily accessible.
Hardware designed for computer vision and analysis is more widely available.
New algorithms like convolutional neural networks can take advantage of the hardware and software capabilities.
The effects of these advances on the computer vision field have been astounding. Accuracy rates for object identification and classification have gone from 50 percent to 99 percent in less than a decade — and today’s systems are more accurate than humans at quickly detecting and reacting to visual inputs.
Hear why Georgia-Pacific chose SAS
Computer vision resembles a jigsaw puzzle
Computers assemble visual images in the same way you might put together a jigsaw puzzle.
Think about how you approach a jigsaw puzzle. You have all these pieces, and you need to assemble them into an image. That’s how neural networks for computer vision work. They distinguish many different pieces of the image, they identify the edges and then model the subcomponents. Using filtering and a series of actions through deep network layers, they can piece all the parts of the image together, much like you would with a puzzle.
The computer isn’t given a final image on the top of a puzzle box — but is often fed hundreds or thousands of related images to train it to recognize specific objects.
Instead of training computers to look for whiskers, tails and pointy ears to recognize a cat, programmers upload millions of photos of cats, and then the model learns on its own the different features that make up a cat.
Computer vision in today’s world
From recognizing faces to processing the live action of a football game, computer vision rivals and surpasses human visual abilities in many areas.
Deep learning and computer vision
How does deep learning train a computer to see? Get an introduction to deep learning techniques and applications and learn how different types of deep neural network models are used for computer vision.
Another set of eyes with computer vision
Georgia-Pacific embedded computer vision in day-to-day manufacturing operations to capture and analyze image data. By constantly monitoring for anomalies, the technology helps solve problems with quality and safety and boosts efficiency.
The Batting Lab
The Batting Lab combines AI, computer vision and IoT analytics with baseball to help kids improve their swings and their data literacy. Sensors and cameras use object detection to collect data points and process footage, creating real-time recommendations to batters.
Computer vision is one of the most remarkable things to come out of the deep learning and artificial intelligence world. The advancements that deep learning has contributed to the computer vision field have really set this field apart.
Learn about the multidisciplinary field of data science
Computer vision for animal conservation
Learn how a computer vision model designed to analyze animal tracks works. Can the computer be trained to see a footprint much like a native animal tracker would? See how the computer processes the different layers of information to determine the animal and its sex. In this video, Jared Peterson, Senior Manager of SAS Advanced Analytics R&D, shows how neural networks are the science behind computer vision.
Seeing results with computer vision
Computer vision users in many industries are seeing real results – and we’ve documented many of them in this infographic. For example, did you know:
- Computer vision can distinguish between staged and real auto damage?
- Computer vision enables facial recognition for security applications?
- Computer vision makes automatic checkout possible in modern retail stores.
From spotting defects in manufacturing to detecting early signs of plant disease in agriculture, computer vision is being used in more areas than you might expect.
Click on the infographic here to see results from retail, banking, health care and more.
How computer vision works
Computer vision works in three basic steps:
Acquiring an image
Images, even large sets, can be acquired in real-time through video, photos or 3D technology for analysis.
Processing the image
Deep learning models automate much of this process, but the models are often trained by first being fed thousands of labeled or pre-identified images.
Understanding the image
The final step is the interpretative step, where an object is identified or classified.
Today’s AI systems can go a step further and take actions based on an understanding of the image. There are many types of computer vision that are used in different ways:
- Image segmentation partitions an image into multiple regions or pieces to be examined separately.
- Object detection identifies a specific object in an image. Advanced object detection recognizes many objects in a single image: a football field, an offensive player, a defensive player, a ball and so on. These models use an X,Y coordinate to create a bounding box and identify everything inside the box.
- Facial recognition is an advanced type of object detection that not only recognizes a human face in an image, but identifies a specific individual.
- Edge detection is a technique used to identify the outside edge of an object or landscape to better identify what is in the image.
- Pattern detection is a process of recognizing repeated shapes, colors and other visual indicators in images.
- Image classification groups images into different categories.
- Feature matching is a type of pattern detection that matches similarities in images to help classify them.
Simple applications of computer vision may only use one of these techniques, but more advanced uses, like computer vision for self-driving cars, rely on multiple techniques to accomplish their goal.
Featured capability for computer vision
SAS® for Machine Learning and Deep Learning
This SAS capability supports clustering, different flavors of regression, random forests, gradient boosting models, support vector machines, sentiment analysis and more, in addition to deep learning. An interactive, visual pipeline environment presents each project (or goal) as a series of color-coded steps that occur in a logical sequence.
Recommended reading
-
Article
Unlocking a strategic approach to data and AIAI is only as good as the data that powers it – this is a fundamental truth about data and AI that defines the limits of what’s possible with artificial intelligence. It may seem surprising, but it's rarely a bad algorithm or a bad learning model that causes AI failures. It's not the math or the science. More often, it's the quality of the data being used to answer the question.
-
Article
What are chatbots?Chatbots are a form of conversational AI designed to simplify human interaction with computers. Learn how chatbots are used in business and how they can be incorporated into analytics applications.
-
Article
Machine learning, data science and AI meet IoTIn this video, Kirk Borne and Michele Null discuss the intersection of machine learning, AI and data science with IoT data and analytics.