Post a Job Join The Guild
Contact Us   |   Sign In   |   Register
Features
Blog Home All Blogs
Search all posts for:   

 

View all (293) posts »
 

Welcome To The Machine - Getting Started With Machine Learning In Media and Entertainment

Posted By Jade McQueen, Tuesday, September 18, 2018

The Media and Entertainment industry, specifically film and TV, has experienced incredible growth over the past few years, with new streaming distribution platforms and a bigger-than-ever global audience eager to consume content on multiple devices. New ways of working, a proliferation of applications and devices, and new types of business processes have resulted in more content and more formats than ever before. With the explosion of content creation comes the challenge of keeping production streamlined, on budget and secure. Amid mounting concerns around data protection and cyber security, there are potential threats of costly hacks from a brand reputation, IP protection and monetary perspective.

To be able to address the new paradigm of production, content creators need to modernize their technology stack in order to digitize business processes. It is vital for production teams to stay focused on what matters most—creating award-winning shows and features. Enter machine learning and artificial intelligence. These four words are going to affect every type of business and are not going away. The automation of manual business processes and gleaning insights from the data and information we create are top priorities for any company that wants to stay relevant in the digital age.

 

WHAT IS MACHINE LEARNING?

The media and entertainment business is no stranger to innovation, as cutting-edge technologies have led (and are continuing to lead) to the creation of exceptional storytelling and experiences. With machine learning, we can train technology to automate and power simple, repeatable workflows from casting to talent agreements. Given the size of the administrative data set generated by the demands of production management, there is an opportunity to apply artificial intelligence and machine learning to this ever-growing, massive amount of information. This opportunity frees up time and bandwidth for creators, transferring the “busy work” to computers. The more content you have access to, the more opportunity to train the technology. Production companies with large libraries and archives of footage, images, VFX, posters and trailers are perhaps best positioned to start training machine learning for automation. 

 

HOW DOES IT WORK?

Many best-of-breed technology providers, including IBM, Microsoft, Google and many more already offer artificial intelligence and machine learning services that can be applied to manual tasks or business processes.  At Box, we are able to leverage the machine learning capabilities from these companies by integrating them into our intelligence offering, Box Skills, and applying the intelligence to unstructured content in Box. The ability to apply these technologies to label objects and images, convert speech-to-text transcripts, and deterct faces in videos are just a few of the ways that we have started to apply some of the practical applications for artificial intelligence and machine learning.

 

What do we mean by intelligence?  And what will the technology “learn?”

Image recognition has the ability to detect individual objects and concepts and can recognize text in image files. Imagine if during pre- and post-production, you never had to add tags manually to photo repositiories from shoots, productions and release parties. All of those images would automatically have tags recognizing characters, text and faces.

Audio Intelligence has the ability to transcribe and identify key topics in spoken audio files, making recordings for auditions and trainings easily searchable by topics or even single words within each file.

Video Intelligence can transcribe and identify key topics for speech and detects individual faces, similarly to image recognition, as they appear in video files.  So you could instantly pull up an archive of past productions that mentioned “California” within the file.

 

How can it help producers and production?

Machine learning is coming to a critical point, as producers shift to new platforms for content distribution, expand revenue streams by monetizing existing catalog and new content, and look to eliminate content silos. What if technology could learn to tackle some of these tasks?


GET GREENLIT FASTER, LAUNCH BETTER

Companies like Cinelytic, a machine learning-driven software as a service platform, is empowering entertainment industry professionals to make faster and better-informed decisions around packaging, financing, producing, distributing and marketing of their content. For clients, Cinelytic provides comprehensive data reporting, predictive analytics, risk and project management tools in an integrated, easy-to-use online system. Using these insights, the system allows producers to develop, produce, finance and market content that will resonate with the audience. For example, producers can now prepare and forecast a business plan for a film project to share with potential investors. While the story remains king and key to success, machine learning can deliver insights on what it takes to create a blockbuster and host a differentiated audience experience.

 

CONTRACT & BUDGET MANAGEMENT: DOCUMENT TEXT EXTRACTION

 Contract management is a time-consuming, complex process that involves numerous document types and countless internal and external teams. Production companies can streamline their end-to-end process by leveraging automation to extract text fields from documents, depending on what information the organization needs. Machine learning is the backbone to this new, streamlined workflow. Technologies like optical character recognition (OCR) and natural language understanding (NLU) are applied to content, creating instructions for data that needs to be pulled from documents. With OCR technology, producers working to on-board new talent can automatically pull required information from a scanned copy, photo or PDF of the contract, such as the talent’s name, agent, lawyer, contract signature date or renewal date.

 

FREELANCE AND CREW MANAGEMENT: DOCUMENT CLASSIFICATION

 It takes a village for production teams to deliver the final cut. Maintaining the coming and going of freelancers and contractors for projects is difficult to organize and it’s hard to ensure that staffing is within budget.

With machine learning, organizations using outside crew and staffers can automatically assign a unique level of data classification to documents, including employment agreements and tax forms. This can be set up to start with on-boarding new contractors and  maintained during their tenure and through the closeout of the project.

For example, when on-boarding new crewmembers, production is likely to run into repeat contractors and staffers that have worked on other projects. Instead of searching through thousands of old records and files to find previous hire information, such as work history and hourly rates, machine learning can tag crew profiles with specific keywords. This allows production teams to easily find the information they need to finalize authorizations and move forward with their hire.

Machine learning can also be taught data retention policies, where specific information included in documents triggers the type of classification that should be applied, where the information should be stored and the length of time an organization needs to retain it for compliance requirements.

 

DIGITAL ARCHIVING AND WORKFLOW MANAGEMENT: IMAGE, AUDIO AND VIDEO LABELING

 Consider a production company’s digital archive. How many thousands of images, footage, posters, trailers and promotional materials exist in these libraries? All of these assets can and should be cataloged, but with the help of computers instead of individuals poring over pages and pages of data.

Whether during pre- or post-production, marketing departments across the industry can use machine learning to automatically recognize specific objects, characters, text and/or faces within the digital archive files. Agents can also use video facial recognition to filter through their talent’s clips and compile highlight reels, instead of digging through loads of content and files.

In the end machine learning is simply about driving more efficiency with intelligence from content and information that already exists within an organization. With all the content and experiences brought to audiences around the world, the entertainment industry is ready to embrace the power of machine learning technology. The applications for applied intelligence are endless, and we’re only getting started.

 

Jade McQueen began her career as an A&R executive at DreamWorks and Interscope Records before transitioning to film and TV. Her love of innovation and technology, coupled with a desire to bridge the gap between entertainment and tech, led to her current position as Senior Managing Director for Media & Entertainment at Box, where she oversees the company’s media and entertainment strategies globally.

This post has not been tagged.

Share |
Permalink | Comments (0)
 
ABOUT THE PGABECOME A MEMBERPRODUCERS CODE OF CREDITSPGA AWARDSPRODUCED BY CONFERENCEPRODUCED BY MAGAZINE