Computer vision (CV) is a field of AI in which computers try to interpret and understand the visual world using images from cameras and videos and deep learning models to accurately identify and classify objects.
Nowadays, images are everywhere; for example, mobile phones, CCTV, YouTube, social media and, of course, the internet (more than 3 billion images are shared online every day). To be able to index and search these we need to know what the images contain. Traditionally, this was done manually by the uploader. But to streamline this process we want computers to be able “see” images and understand what’s in them.
For us humans, this is straightforward; we can describe images, we can summarise video and we can recognise objects and people - we want similar capabilities in our algorithms so we can unlock further value from images and videos.
Enter, CV.
The goal of CV is to discover and understand the content of digital images. Usually, this involves developing techniques that try to reproduce human vision - no easy task! In fact, it’s an extremely challenging task for a number of reasons. For example, any person or object may be seen from any orientation, in any lighting condition, with any type of occlusion from other objects, and so on. A true CV system must be able to “see” in any of an infinite number of scenes and still extract something meaningful. Furthermore, there are very few working and comprehensive theories of brain computation - scientists are not unanimous on how the brain and eyes process images, so it’s difficult to say how well the algorithms used in production approximate our own internal mental processes.
Nevertheless, there have been extraordinary achievements in the field thanks to advances in AI and, in some cases, AI has been able to surpass humans in CV tasks related to detecting and labelling objects.
For example, some systems have reached 99% accuracy - making them more accurate than humans at quickly reacting to visual inputs and hence there are now robust solutions available for:
- Face recognition
- Optical character recognition (OCR)
- Machine inspection
- Retail (e.g. automated checkouts)
- 3D model building (photogrammetry)
- Medical imaging
- Automotive safety
- Surveillance
- Fingerprint recognition and biometrics
- Healthcare
- Self-driving cars
CV has impacted many industries; Retail, Security, Automotive, Healthcare, Agriculture, banking and industrial to name a few. Let’s take a closer look at some of these.
In security CV can be used to recognise faces to verify the identity of an individual and unlock devices such as phones or in social media applications to offer suggestions to automatically tag users in an image or video. In self-driving cars CV and AI examines the surroundings to identify objects, pedestrians, traffic signals, and more, on the road. It alsos helps braking systems and prevent accidents by measuring distances between vehicles. Finally, in healthcare together Computer vision and AI can be use to automate or assist in tasks such as detecting areas of interest in scans.
There are many specialised solutions and techniques and there is no one-size-fits-all solution - but computer vision is here - it’s awesome! - and it’s only going to get better.