What does Computer Vision mean?

Computer vision is the branch of AI that automatically analyzes images, video, and scans. Such systems recognize objects, people, text, or movement within them and output the result as data that can be processed further. Unlike plain image editing, computer vision does not alter the image itself but derives a verifiable statement about its content.

An image enters the model as a matrix of numbers, with one value per color channel and pixel. Convolutional neural networks slide filters across this matrix, first detecting edges and, in deeper layers, shapes and whole objects. Vision transformers instead split the image into tiles and weight how those tiles relate to one another. Quality hinges on annotation, since every training image needs a verified label describing what it shows.

Computer vision pays off wherever large volumes of image material need to be checked against the same criterion, over and over. It is used, for example, to assess the condition of roads and utility networks, for quality control in manufacturing, and to read out forms and receipts. Where a single high-risk case is at stake, review by a qualified person remains the foundation.

The advantage over manual visual inspection lies in a consistent standard. Every image is judged by the same criteria, regardless of time of day or fatigue. Because the model outputs a confidence score for every match, a threshold can be set and everything below it routed for a second look.

As soon as people, license plates, or faces appear in an image, the GDPR applies and the purpose must be defined in advance. Blurring or masking directly at the point of capture keeps the data volume small and prevents any personally identifiable image from being created in the first place.