Maximize the potential of AI for media workflows with Accurate.Video from Codemill
Authored by Neil Anderson, Project Consultant at Codemill. The content and opinions in this post are those of the third-party author and AWS is not responsible for the content or accuracy of this post.
Over the last year, consumer demand for video content has exploded. According to Research and Markets, the global video streaming software market is expected to grow from USD 6.1 billion in 2020 to USD 15 billion by 2025. New SVOD services have launched to meet this growing demand, including Disney+, Apple TV+, HBO MAX, Peacock, Discovery+, and soon Paramount+ from ViacomCBS. Meanwhile, existing services, like Netflix, are looking for new ways to expand and to enhance their offerings. More competition means that providers must add to their offerings, producing higher-quality (4K HDR at a minimum) and adding more engaging video content. At the same time, content owners also need to keep subscription costs affordable to attract and retain viewers in an increasingly crowded market.
With this increase in demand, broadcasters and studios are looking to maximize workflow efficiencies and reduce capital costs, while continuing to deliver higher quality video content to viewers. A “cloud first” strategy is an obvious way to pay only for the capacity they use when they use it, and being able to scale resources up or down as required. At the same time, broadcasters and studios are increasingly looking at new ways to optimise their production workflows and content supply chain. Could Artificial Intelligence (AI) be the ultimate solution?
AI technology has rapidly improved in recent years. Specifically, machine learning and pre-trained analysis services, such as Amazon Rekognition, expand the possibilities of AI beyond what was possible just a few years ago. Amazon Rekognition makes it easy to add video analysis into media workflows, using proven, highly scalable deep learning technology that requires no machine learning expertise to use. This has the potential to enable huge efficiencies, which ultimately saves both time and money. Examples include creating automated transcriptions and subtitling, or enabling celebrity recognition to quickly extract footage for news stories, or to verify brand and image rights.
AI services can automate the identification of niche or industry-specific content. Organizations can use Amazon SageMaker Ground Truth to identify specific racing drivers, car models, or animal breeds for example. Ground Truth is a fully managed data labeling service that makes it easy to build highly accurate training datasets for machine learning.
AI can handle some of the more labor intensive, mundane, and repeatable processes, such as automatic shot logging, transcriptions, and segmentation of content for ad breaks. This allows creatives to spend more time doing what they do best, while the media operations team can work more efficiently.
Pairing Accurate.Video and Amazon Rekognition
Accurate.Video is a web-based video platform created for remote broadcast, post-production, and media professionals. Accurate.Video Validate enables content operations teams to frame accurately Quality Check (QC) video, with discrete audio tracks and multiple subtitles. It can also verify time-based metadata, all using a standard web browser, with extensive keyboard shortcuts.
Validate integrates with broadcast auto-QC reporting software, providing operators with sophisticated metadata timeline visualization so they can easily check warnings and error flags. Users can add frame accurate manual markers, ranges and annotations, or set a custom asset status, such as passed, failed, or warning.
Accurate.Video Validate includes API integration with Amazon Rekognition to automate some of the otherwise labor-intensive processes associated with quality control of video content. There are a whole host of possible use cases for AI, each of which can deliver huge efficiencies. These are just some of those possible with the API integration between Accurate.Video Validate and Amazon Rekognition:
1. Identification of inappropriate content
Finding inappropriate content to meet compliance and regulation standards is challenging and labour intensive, but is also extremely important. Standards can vary depending on where and when that content is delivered. Adhering to different regulations at certain times of the day or in different regions can be very challenging, especially for multinational broadcasters. It is easy for a person to miss something, especially if they are reviewing a lot of content with tight delivery times. AI can quickly flag potentially inappropriate content to complement human compliance processes.
2. Finding shot changes
Users can quickly find shot changes on edited footage using Accurate.Video. This helps content operations teams to quickly determine where to place ad breaks for commercial broadcasters, or where to add “binge markers” on VOD assets, when title end credits appear.
3. Transcript creation
AI can rapidly create draft audio transcripts, a process that a human can only do in real time. Someone still needs to review and refine these for accuracy once generated. AI transcription is already highly accurate, often requiring only slight alterations. Of course, this can depend on regional accents and background noise that can be challenging for AI to decipher, but the technology continues to rapidly improve. Further retraining of the machine learning model can have a positive impact because you can teach it to “pronounce” and better recognize some of the more challenging accents and dialects. Users can also train AI to recognize terminology specific to the industry or type of video content it analyses.
4. Staying organized
Sometimes just finding content contained within a growing archive can be extremely challenging. Using Amazon Rekognition with Accurate.Video lets media organizations detect faces, celebrities, objects, logos and branding, human sentiment, and even the presence of on-screen text or captions. Time-based metadata makes it easy for broadcasters and studios to repurpose their existing content, whether it’s for historical event documentaries, or to create new and original productions from previously shot material.
The future of AI for media
In each of these use cases, it is clear that AI is only part of the story. Critical human decision making is still vital for ensuring quality, compliance, and consistency. However, AI gives broadcasters, studios, and media creatives a helping hand. AI creates significant efficiencies for media production workflows. As the technology continues to evolve, especially when combined with custom AI training, the use cases only continue to grow.
Accurate Video.Validate is available on AWS Marketplace. Learn how to get started using Accurate.Video Validate from the AWS Marketplace using our step-by-step setup guide, or contact us to learn more about using the power of AI to increase automation in your media workflows.