Amplifying Human Content Expertise with Real-World Machine-Learning Workflows

Human-led content classification and enrichment have long been the most impactful yet most expensive form of content workflow operations. Large content library owners often find that they have irreplaceable content expertise concentrated in only a few contributors. They become one of the critical gating factors in effective library content utilization. — Recent advancements and productization of innovative AI technology empowers accumulated human expertise and liberate teams from tedious manual activities, enabling much higher productivity and creativity. By applying the latest AI Computer Vision, Natural Language Processing, Machine Learning, and Video Analytics, a content library can be quickly transformed from a pile of ingested media and tape files to rich content that is fully annotated, searchable, and enhanced for efficient user consumption. — Augmenting the content processing workflows in commonly used tools such as a Media Asset Management system with AI functionality brings an immediate benefit to the end user and abstracts the AI technology complexity. — This paper provides an overview of the AI capabilities applicable to the Media and Entertainment industry. It outlines challenges introduced by the new technology capabilities and best practices to overcome them. — In this paper, we take the user through the approach, steps, and processes of taking a timeconsuming, novel content application task usually performed by a human expert, adapt that task to an AI challenge, build and deploy the necessary ML model, then apply it to content as a part of ongoing content operations using familiar asset management systems.

Published

2022-10

Content type

Original Research

Keywords

Content discovery, annotation, AI, ML, MAM, DAM, media management, automation

DOI

10.5594/M001968

ISBN

978-1-61482-963-8