Microservices architecture for content-based indexing of video shots

Abstract

Three different content-based video indexing microservices dedicated to index video shots for the needs of the IMCOP Content Discovery Platform are presented in the paper. These three services as well as numerous others cooperate with each other within the IMCOP platform to describe, enrich and relate the multimedia data regarding their audio, textual and visual content. Owing to the analysis they perform, the IMCOP platform can discover, recommend and deliver the personalized multimedia content to various IMCOP’s prospective recipients. As these recipients may also require the personalized video content, services, as e.g. the presented ones, designed respectively to discriminate between characters in videos as well as text- and speech-based indexing of video shots, are absolutely essential. Goals of these services, their approaches and how they comply with objectives of the IMCOP’s microservices architecture are carefully presented in the paper. Research procedures and the results of examinations that have been carried out to verify their pretty high accuracies are also reported and discussed.