AI Industry

AI in the Media and Entertainment Industry

The Media and Entertainment industry is a cornerstone of contemporary human culture, delivering films, TV shows, advertisements and more in a multitude of languages across a wide variety of devices.

The Media and Entertainment industry is a cornerstone of contemporary human culture, delivering films, TV shows, advertisements and more in a multitude of languages across a wide variety of devices. A PwC report predicts total M&E revenue will reach US$2.2 trillion in the next three years. The industry’s growth rate however has lagged, and slipped by 0.2% in 2017, prompting many companies to turn to AI technologies to boost business.

Four Major Categories Where AI Is Involved

With the breakthroughs in machine learning, many intelligent products have made the leap from sci-fi movies to the home. Superhero Ironman’s virtual helper JARVIS (Just A Rather Very Intelligent System) for example is echoed in smart assistants such as Alexa and Google Assistant, which may not catch criminals but can perform a range of practical tasks via connected household devices. NVIDIA meanwhile used VR technology to create a Holodeck similar to one in the sci-fi series Star Trek.

AI technologies are also being applied in filming, visual design, post production etc.

Current AI applications in the M&E industry are mainly in four categories: Marketing and Advertising, Service Comprehension, Search and Classification, and Experience Innovation.

Marketing and Advertising

The marketing and advertising sector includes visual design, film promotion and advertising. A machine learning algorithm trained with data such as text, stills and video segments can extract language, objects and concepts from its training resources and suggest marketing and advertising solutions to improve efficiency. Such a system can work as an assistant or even a content creator.

Morgan poster.jpg
Poster for the 2016 science fiction horror film Morgan

Alibaba’s Luban is an AI designer that can create banners thousands of time faster than a human designer. On China’s online shopping extravaganza “Singles Day” in 2016 Luban generated some 8000 different banner designs per second and 170 million banners in total. The record output would of course be impossible for human designers to process in one day. On Singles Day 2017 Luban raised its one-day record to a staggering 400 million banners.

IBM used their AI system Watson to help 20th Century Fox create a trailer for the horror movie “Morgan.” The research group trained the AI system to analyze and classify input “moments” from visual, audio, and other composition elements in 100 horror movies to learn what kind of “moments” should appear in a standard horror movie trailer. Watson needed just 24 hours to create a six-minute movie trailer that may have taken a human professionals weeks to produce.

Albert Intelligence Marketing’s AI marketing platform accelerates the marketing process using predictive analytics, machine learning, NLP and computer vision technology. The Albert platform can perform audience targeting, customer solution making and generate autonomous campaign management strategies. Albert says companies using the platform report a 183% improvement in customer transaction rate and 600% higher conversation efficiency.

Personalized Services

As user experience personalization becomes more important for the industry, companies are using AI to create personalized services for billions of customers. These include for example recommending content that fits users’ personal tastes while they are browsing a video site or shopping online; and optimizing video definition and fluency for users with different internet speeds and bandwidth.

Netflix’s content recommendation got a boost in May 2016 when the company launched Meson, an intelligent workflow management and scheduling application. This AI system automatically manages the various machine learning pipelines that provide video recommendations. According to the Netflix 2016 annual report, there are 93 million global users streaming over 125 million hours of TV shows and movies per day on the platform. Predicting which shows will attract users’ interest is a key component of the Netflix model.

AI technology is also being applied to optimize video fluency and definition. Slow Internet connections and bandwidth limits can be a problem for streaming services in developing nations and for mobile device users. Netflix collaborated with the University of Southern California and the University of Nantes in France to develop a new machine learning methodology called Dynamic Optimizer which can compress video without degrading image quality to ensure a smooth and high quality streaming experience for its customers, whether they are in India or in Japan.

Search and Classification

The Internet hosts countless media works. Video, audio and text can all be transformed into a digital copies which can be stored and spread so easily that it is getting increasingly difficult for people to find exactly what they want online. AI is helping optimize the accuracy of search results. Computer vision technologies meanwhile are also enabling content producers to better manage visual content and accelerate the media production process.

Advancements in machine learning technology have enabled Google to augment the world’s leading search engine in multiple ways. One is in image searching. Rather than typing in keywords and checking returned images, users can upload a sample picture to Google Image, which uses image recognition technology to identify image features and search for similar pictures. Another advanced application involves selective link-building. Google applies AI to position ads appropriately — for example so a cat food ad appears in a pet-related website, but a bacon cheeseburger promotion will not appear on a site for vegetarians.

ClarifAI is an AI startup focusing on computer vision technology which partnered with Vintage Cloud to deploy AI on a film digitalization platform. By using ClarifAI’s computer vision API, Vintage Cloud successfully accelerated the progress of movie content classification and categorization. It used to require dozens of hours for humans to recognize and manually classify objects in a movie. AI can do a better job in much less time.

image.png
ClarifAI’s image recognition API

Experience Innovation

In the past, papers and books were the main medium for words and images. The introduction of film and TV brought us into the dynamic new world of moving pictures. Now, AI is heralding a new age of immersive experience for visual content. This technology includes Virtual Reality (VR) and Augmented Reality (AR). With machine learning algorithms and computer vision technologies, developers can build complex and holographic scenes within a pair of goggles. This opens up a brand new market.

VR gaming is one of the first areas that comes to mind, and this is where companies like HTC Vive, Samsung Gear VR, Oculus Rift, etc. are focusing their efforts. Various type of headsets have been introduced. Combined with motion sensing games, VR gaming innovation has become a hot market that shows no sign of slowing down.
pick-a-best-VR-Headset-india.jpg
Intel is now getting into the immersive experience industry. With the application of deep learning and computer vision technology, Intel has become a visual content provider emphasizing Virtual Reality content. Supported by AI algorithms, Intel True VR Technology can perform every piece of a scene with pixels in three-dimensions.

Using the tech, fans can also watch sports in holographic view. Intel demonstrated this in their widely viewed VR game broadcast of the NFL 2018 Super Bowl. Intel partnered with the International Olympic Committee to broadcast the 2018 Winter Olympic Games as 360-degree video content. With a VR headset or even just a smartphone, fans and families could experience the action from the POV of an athlete.


Analyst: Victor Lu| Editor: Michael Sarazen

0 comments on “AI in the Media and Entertainment Industry

Leave a Reply

Your email address will not be published.

%d bloggers like this: