Image content search is the capacity for software to recognize objects in digital images and return a search engine results page (SERP) based on a user query. Should the user request a particular breed of dog, for example, the software would analyze indexed images to identify any examples of that breed. In contrast, traditional image search looks for keywords in content associated with images through text or meta-tags.
Image content search opens up many new possibilities for smart photo libraries, research, targeted advertising, interactivity of media and accessibility for the visually impaired.
Although humans recognize objects with little effort, computers have difficulty with the task. Software for image content search requires deep learning computers with neural net processors and a lot of processing power for the compute-intensive task. Image content search engines are often trained on millions of tagged images in guided computer learning.
One early application of image content search is Facebook's Lumos computer vision platform. Lumos was originally designed to identify what is in a picture and what is happening in it and describe the image for visually-impaired users.
Image content search is also known as content based image retrieval (CIBR) or query by image content (QBIC).