The recent surge of possible media platforms, especially mobile devices such as tablets and phones, alongside the increased pervasiveness of internet access, has resulted in an increased requirement for on-demand content. Consequently, the media production industry worldwide has had to rapidly adapt to remain fit for purpose. This has necessitated the development of cost-effective media workflows which are fast, flexible and accommodate various source and play-out formats. This onslaught of new content challenges more traditional practices for quality assurance and Media Asset Management (MAM), which depend on rich metadata and data integrity.
Current enterprise solutions to these challenges, for reasons of expense, implementation complexity and their proprietary nature, are often beyond the reach of smaller creative endeavours. They are usually only accessible to the larger public and private production and broadcast companies. The RECAP (Real-time Content Analysis and Processing for Agile Media Production) project has been established to redress this balance. It is doing so by building on recent research from automatic real-time content analysis and processing, to bring to market an affordable, integrated, scalable, open commercial software solution.
The RECAP team recently demonstrated a working prototype at NAB 2017, the world''s largest broadcast tradeshow. The demonstration showed how analysis workflows can be managed using the onCore software from Tools On Air, with the results viewable on a web-based user interface. Summarising the advantages, the CEO of one of the consortium members (NMR) Neil Anderson says, ‘Customers with live video workflows or existing archived video are never going to have time to watch, analyse, and make decisions about all of their content.’ The interface facilitates their analytical procedure.
Introducing the proof of concept
The RECAP software platform works through modules which contain algorithms that can be integrated into real-time media analysis workflows. These algorithms are designed to automatically extract descriptive and technical metadata, and can be plugged into MAMs, allowing content producers to analyse content, in real-time, from live broadcast-quality video. The same approach also caters for batch processing of existing archive content, as well as being flexible enough to be deployable on-site, at a data centre or as a cloud-based service.
Ensuring a market-ready final product
The final RECAP product will deliver broadcast-quality video with real-time metadata streams, available for instant review in a web browser based timeline. It will provide analysis and further processing in commercial MAM systems, workflow orchestration tools, and transcoding software by means of open and documented APIs. The descriptive analysis of video and audio content will include objects, logo and face detection, along with content duplication.
Additionally, to help with quality control, the technical metadata will also automatically detect impairments, such as dropped video frames, with the possibility of applying enhancements when problems around quality issues such as sharpness and noise, are identified. As well as the usage of the system for archives, additional functionality under development includes the identification of on-screen text (OCR) and, as an aid to content search, automatic speech-to-text (ASR).
To ensure that RECAP addresses real world scenarios and meets industry standards, the project team is working through a stakeholder board of major broadcasters, news organisations, and content owners, who provide the necessary guidance. The team is also set to undertake a series of demos at major trade shows, conferences and partner events over the coming months.
For more information, please see:
project website