Video content is a primary business asset of the thriving UK and global creative industries, and represents a substantial proportion of today’s “big data” deluge. Beyond the creative industries, video comprises a major communications tool, occupying 60% of today’s Internet traffic. However, with millions of users creating, processing and (not always legally) uploading exabytes of video content each week, video remains the least-manageable element of the big data ecosystem. This is because all current methods for high-level semantic description in video require either manual annotating or compute-intensive video decoding & processing. Delivering cost-effective and meaningful video search has therefore proved to be an insurmountable problem. This project takes a novel insight: that a hidden source of coding-related metadata already exists within modern compressed video file containers, and it is sufficient for automated tagging and visualization. Tapping into this “hint” metadata enables video streams to be analyzed with 3~6 orders increase in speed and decrease in cost, enabling exabyte-scale video datasets to be newly-discovered and analysed over commodity hardware.
Video dominates global business and consumer Internet traffic and continues to grow (2012: 52%, 2017: 67%, Cisco VNI), demonstrating the massive impact of visual media production and media distribution services. Investments in these services can be severely compromised by poor visual quality, caused by mediocre encoding tools, lack of encoding and streaming know-how, and failure to support the nuance of today’s platform diversity (from HDTV, soon UHDTV, through to mobile displays). The REVQUAL project delivers three inter-related innovations which sustainably enable visual quality excellence to be achieved comprehensively. These include: (i) new approaches combining machine learning and existing state-of-the-art visual quality metrics, delivered through: (ii) a new ‘web crawler’ service for automated video quality optimization within video production and distribution services, supported by: (iii) a visual quality human assessment database, optimized to professional media content, creating an open dataset, incorporating the emerging principle of “social visual quality assessment”. The project sets a new baseline for quality, enriching the value of all services.
The VIDAS project investigates the feasibility of using visual quality metrics to assess
content similarity in an automated manner. This has several applications, including: piracy
detection (which could include the automation of take-down notices), de-duplication of stored
files, version detection, malicious or sensitive content detection, and ‘pre-flight’ inspection of
materials for digitization or re-mastering. The project builds on a successful earlier TSBsupported
BAFTA-UCL project (VQ-INDEX), which innovated a new mechanism for
assessing visual quality in comparison with an original source. The VIDAS project will
reverse this paradigm, and (with the addition of correlators, scene-cut and video resolution
detectors), determine whether discovered content was derived from or is similar to a copyright
holder’s original. The objective during the proof of concept study is to: a) investigate the
feasibility of applying the approach of using visual quality metrics for similarity
identification, and b) confirming the technology can be used to address key market
requirements across a range of digital assets and requirements. Overall, this provides
significant automation to processes currently performed by human viewers.
The VisibleRights project will reverse the deeply-ingrained convention of rights clearance as a post-sale activity in the arena of media repurposing. Reversing the trend from reactive to proactive rights management opens up new collections, engages new content sources and substantially raises the volume of media available for licensing, discovery and reuse. The project will combine use of both commercial and Creative Commons licensing, and provide a coherent and unified route to market for content. In parallel, the project creates and disseminates an evidence base to assess the correlation between open access and commercial traction, which the project partners believe to be positive. Media types including moving images and print publications will be included, enabling a granular level of content licensing, from a few seconds of footage or a single journal article, in a common mechanism. In addition to supporting a new cross-collection licensing portal for 70 members of London Screen Archives and other inaugural users, the project will deliver a rights clearance toolset and a persistent rights-aware dataset available for query.