TrackIt Uses Machine Learning to Automate Postproduction Tasks for Fuse Media on AWS

AWS Partner TrackIt helped Fuse Media to build a custom video player application designed for content curation and review, a solution that saves up to 10 hours of manual work per project in the postproduction process. Fuse Media is a Latino-owned global entertainment company with a portfolio of brands. These include a suite of free ad-supported streaming television (FAST) channels focused on diversity, equity, and inclusion; a subscription service; and two linear cable channels. TrackIt and Fuse Media use AWS machine learning (ML) and artificial intelligence (AI) solutions to detect questionable content in media files. A web-based user interface also greatly simplifies the workflow for editors.

Fuse Media was looking to streamline its postproduction processes, which often involved the time-consuming task of detecting and filtering out questionable content. Editors and other team members involved in the editing process had to exchange emails and review changes often, which was inefficient and frequently caused delays. Following the recommendation of Amazon Web Services (AWS), Fuse Media began to work alongside AWS Partner TrackIt to develop a solution to help make the editing process more efficient. Using ML and AI solutions from AWS, TrackIt designed a solution that helped Fuse Media fulfill its goals of saving time in the edit room, curating content to meet certain guidelines, and creating a user-friendly interface so that editors can work more efficiently.

As a multiplatform entertainment company, Fuse Media creates and distributes purpose-driven stories created by—and for—diverse young adults. The Fuse Media family includes Backstage, Shades of Black, OUTtv Proud, its in-house production arm Fuse Content Studio, and a growing branded-content and live-events business.

As a media and entertainment company, Fuse Media handles huge volumes of content daily. During postproduction—the editing that occurs after the completion of filming—one of the company’s primary challenges is the need to consistently detect and filter out questionable content. This content might include graphic imagery, nudity, or swear words. “Like most organizations, we’re constantly battling resourcing and trying to find ways to work more efficiently,” says Bejon Parsinia, senior vice president of technology at Fuse Media. “We found that the specific workflow we had was consuming a lot of human labor to accomplish. We figured there had to be a better way and started to search for solutions.”

Fuse Media contacted AWS to talk about the concept of using ML and AI to automate the time-consuming tasks of filtering content. AWS suggested that Fuse Media work alongside AWS Partner TrackIt, which provides cloud-management products and professional services with special expertise in media and entertainment workflows. In only 3 months, TrackIt and Fuse Media built a pipeline so that editors could use ML and AI to streamline and accelerate postproduction.

Credit: Fuse Media

Fuse Media was interested in building a solution around Amazon Rekognition, which automates image and video analysis with ML. Amazon Rekognition helps extract information and insights from images and videos through pretrained and customizable computer vision, a subset of AI that trains computers to understand and automate the tasks of a human visual system. “TrackIt came onboard to help us realize what other components would be needed to make the overall workflow a reality,” Parsinia says. “We had the conceptual idea of what we wanted to accomplish and how to go about it. TrackIt helped us define a meaningful workflow that we could actually implement using AWS.”

The new workflow begins with a web-based user interface where Fuse Media users upload video to Amazon Simple Storage Service (Amazon S3), an object storage service offering industry-leading scalability, data availability, security, and performance. Fuse Media can migrate data into Amazon S3 from storage in the cloud, on premises, or other applications. The upload of a video asset to Amazon S3 initiates a function of AWS Lambda, a serverless, event-driven compute service that lets organizations run code for virtually any type of application or backend service without provisioning or managing servers. The AWS Lambda function initiates the appropriate media analytics solution, such as Amazon Rekognition or Amazon Transcribe, which automatically converts speech to text. The transcriptions serve as searchable archives, another way the solution helps Fuse Media moderate content. After Amazon Transcribe has completed the conversion to text, another AWS Lambda function pushes the logs to Amazon CloudWatch, a monitoring and observability service built for DevOps engineers, developers, site reliability engineers, IT managers, and product owners. Fuse Media uses Amazon CloudWatch for visibility into the AWS resources it uses, supporting metrics such as the number of transactions, errors, and latency associated with the transcription jobs. This entire workflow pipeline is managed by AWS Step Functions so that the whole process can be built and deployed rapidly.

The TrackIt team also helped Fuse Media to build a custom, web-based video player application specifically designed for content curation and review. The video player dashboard displays a list of ingested files, highlighting occurrences of questionable content. Editing staff can review clips, accept or reject identified markers, and add notes. “I’d absolutely recommend TrackIt to other companies,” says Parsinia. “They’re professional; they understand the inner workings of AWS and a lot of the components that need to come together to build cohesive workflows. They’re not afraid to roll up their sleeves, get their hands dirty, and figure things out. Whenever we’ve had an idea, they’ve had an answer.”

Depending on the type of content, the editing team saves 6–10 hours per project. “Our team is absolutely excited about the workflow that’s been implemented,” Parsinia says. “This not only saves time from an editing standpoint, but there are other teams that are also going to realize time savings from this solution.”

Fuse Media’s success has implications beyond its use case. The AWS team believes that TrackIt’s solution has applications across the media and entertainment industry. “I think that this solution we’ve put together with TrackIt can certainly benefit the media and entertainment community in general,” says Parsinia. “It obviously is serving a need, and I’m pretty sure other people in the media space would want to benefit from a similar workflow.”