What is Computer Vision?
Computer Vision is the science and technology of machines that see. Computer Vision is held as a scientific discipline which concerns theory and technology for building artificial intelligence systems which gather information from images and multi-dimensional data. Artificial Intelligence largely deals with careful planning and deliberation for systems which can perform mechanical actions: such as moving hardware within production lines. This form of intelligent data processing typically requires input data which can only be provided by a Computer Vision system, acting as a vision sensor and producing high-level information about the working environment.
In short, Computer Vision revolves around the automatic extraction, analysis and understanding of useful information from a single or sequence of images. The process involves the development of a theoretical and algorithmic basis to achieve an autonomous visual understanding.
Why is Computer Vision Being used?
Computer Vision is reaching an advanced stage, and is being used across daily life and business operations to conduct multiple functions such as face recognition, identification, verification, emotion analysis, and crowd analytics. Further uses include: warning drivers of animals on the road and pinpointing issues in x-rays. Gaming has also seen Computer Vision being integrated into its systems such as the Xbox Kinect and PlayStation Eye which “sees” and analyses our movement.
Computer Vision is now being used to scan social media platforms to find relevant images which would not be available through traditional searches. The technology itself is far more complex and intuitive, and just like the aforementioned tasks, it requires more than just image recognition, but also semantic big data and analysis.
It has taken computer scientists almost 80 years to get to where we are today and with Deep Learning, Data Modelling and Computer Vision, we are continuously refining the Artificial Intelligence landscape.
Computer Vision in Industry
Computer Vision is an outstanding technology showing major progress in transforming the medical landscape. The extensive use of Computer Vision and medical imaging data is used to provide patients with better diagnosis, treatment and prediction of diseases. Computer Vision exploits texture, shape, contour and prior knowledge alongside contextual information from image sequences, providing 3D and 4D information that helps us build a greater human understanding and medical assurance.
Many powerful tools revolving around image segmentation, machine learning, pattern classification, tracking and reconstruction, are becoming mainstream because of the much-needed quantitative information not easily available by trained human specialists.
To put this into perspective: when a pathologist has, on an average day, 500 slides, each containing hundreds of thousands of individual cells that need to be analysed for cancerous properties, it can be easy to miss a diagnosis. It is considered as an impossible task for humans to do this as effectively as a computer, simply because of the fact that we are not able to carefully analyse every single cell in a feasible and effective manner.
Lidar & Radar
By 2025, 75% of cars on the road will be considered as Smart Cars. Though not yet legal to become fully autonomous, today’s Smart Cars use sensors, Lidar (Light Detection and Ranging), radar, cameras and image recognition systems to “see” the environment around them. Features can range from autonomous driving to safety features such as emergency braking and animal detection through the use of radar. This will undoubtedly change the way we commute in the future and arguably shape FMCGs delivery sector.
It is impossible for humans to physically see heat or gas. In many cases (especially where fires, wild predators and gas leaks are concerned), there are dangers we would want to be notified of before we can see of feel them. Advancements in thermal imaging processes have allowed this technology to not only be built into portable cameras for industrial and consumer users, but also into our smartphones. The anticipation and prevention of dangers are not the only things thermal imaging can be used for. Sporting events are now using this technology to detect mechanical doping in an attempt to keep sports honest.
The identification of a specific object (pedestrians, landmarks and traffic) is made possible by the access to real-time location via GPS and the cloud. For example, Google Photos is able to recognise a differentiate between images of the Petronas Tower and the Tower of London when it tags its user’s photo albums. Geo-location can also be developed to deliver warnings to drivers such as traffic, weather hazards and even when he or she is about to collide with a cyclist (and vice versa) – an added safety feature if the smart car’s lider or radar does not initially detect the instance.
From historical traffic patterns and weather reports, to observing public online behaviours, the information that computer vision can access via the cloud gives it the ability to discern anything from everything. From a business perspective, Computer Vision can be used to track consumer behaviour and learn what ads to display where, when and to whom.
As with human vision, Computer Vision is significantly more than just simply being able to see things. Realistically, it requires a connection to many other data-gathering technologies to be able to produce and deliver realistic and accurate insights which ultimately result in safer methods of transport, smarter homes and optimised businesses.
Sensors which detect changes in environmental mediums (light, air quality, gas and motion) are just a handful of features Computer Vision uses to identify its surroundings. Let us look at today’s smart buildings. They use sensors built into their lighting and temperature systems to detect the movement of people, optimising light and adjusting energy levels whilst becoming ‘smarter’ over time.
Additionally, home monitoring systems not only use motion sensors to detect any security hazards, but combine these cameras and sensors to create a perfect living environment by adjusting lighting, air quality and room temperature on a real-time basis. Moreover, in-store sensors and beacons which work in harmony with cameras, track shopper’s movements whilst cross-referencing them with ‘big’ behavioural data available from the cloud. The objective of this technology within the retail sector is to provide them with information on optimising store layouts and pricing as well as serving coupons to serve their customers in real-time – changing the modern landscape of retail.