Knowledge extraction from web images

Abstract

Given an image of an event or object, key information such as location, description and other attributes may not be ascertained purely by visual inspection. Usually, this image would be searched on the internet through a search engine and after perusing through various websites, the information required could be obtained. The integrity and accuracy of the information obtained can be further enhanced by cross-referencing across multiple resources or websites, unless a reputable or trustworthy website provides the necessary information. Images uploaded onto the internet are often tagged by the uploader which could be deemed useful if the tags provided are comprehensive and accurate. This provides textual metadata which are then used by search engines to provide results when a search is executed. There are various other methods of which an image is associated with other visually similar images or tags such as using colour features that correspond to other images with similar colour profiles [1]. The goal and objective of this project is to mimic the research and information that would be gathered by a human by using artificial intelligence (AI). The program or AI designed in this project would utilise Python programming language.