It is no secret that computer vision is rapidly changing our lives.
Images and videos are an integral part of our everyday lives in countless ways – medical procedures, e-commerce, security, technological interaction and many other fields related to daily activities. Did you know that on Facebook alone, around 350 million images are uploaded every day? And that over 500 hours of video are uploaded to YouTube every single minute?
Hardware and software advances now allow computers to review, analyze and provide meaningful outcomes from images and videos. Nowadays, computers are nearing the replication of the human vision and even surpass it in some respects.
The economic impact of computer vision is growing rapidly. The global computer vision market was valued at 10.6 Billion US$ in 2019 and it is expected to grow at 7.6% compound annual growth rate between 2020-2027.
But what is computer vision?
- Computer vision is a field of computer science that enables computers to replicate the complexity of human vision.
- Computer vision is all about extraction of useful information from images and videos.
- Recent advances in artificial intelligence, deep learning and neural networks have enabled rapid adoption of computer vision and new breakthroughs are happening every day.
- Computer vision includes and enables various tasks, such as detection, pattern recognition, classification, segmentation, and others.
Different types of algorithms are used for image analysis in computer vision:
- Image classification algorithms – Automatically providing a specific label to an image which describes the image’s content.
- Object detection algorithms – while classification algorithms explain what object is visible in the image, object detection algorithms provide information about the object’s location in any given image. These algorithms have endless functionalities.
- Segmentation algorithms reveal which pixels correspond with each object, thus make images easier to analyze, providing additional data about the images and their objects. There are some types of segmentation : semantic segmentation marks all pixels that belongs to a certain type of object with the same classification, while instance segmentation classifies various objects of the same type differently. Region based segmentation uses the edges in the image and the change in spatial characteristic in the image, and many more.
Computer vision is used to solve many problems and affects almost every aspect of our daily routines. Here are some examples:
- Face recognition, for example, has a wide range of applications – building a virtual makeover system, supporting law enforcement (detection and recognition of criminals), increasing security, tagging people in images etc.
- In the medical world, it helps to diagnose a patient faster and more accurately. Tumor detection is based on tissue images, blood flow monitoring, etc. Automatic navigation inside the body helps during medical procedures – and all these procedures rely more and more on computer vision.
- In the transportation world, computer vision is the basis of autonomous vehicles, that depend on it. For a car to drive automatically, a clear understanding of the world around it is required and it is mostly based on images and their correct interpretation.
- In the aerial field, when implemented in drones for example, computer vision helps to navigate, to lock on a target, avoid any obstacles and track changes over time.
- E-commerce has also moved towards digitalization and automated technologies that rely on computer vision. For example, applications that recognize clothing items, clothes suggestions and demonstration of how clothes and accessories will look on you.
- Augmented and mixed reality relies on computer vision capabilities. These applications require elements such as depth and dimensions to place virtual objects in the physical world, these are computed from the image sequences using the various computer vision techniques.
As the world entered a new decade, we can expect to see exciting new innovations and practices that will be based on computer vision.
Milli Peled, VP Marketing