February 6, 2019

Video Marketing and Artificial Intelligence

The capabilities and applications of artificial intelligence have grown by leaps and bounds during the last few years with the help of mainstream computer-chip makers. Aspects of video generation and the related field of computer vision have advanced quickly, thanks to the development of AI software algorithms that are enhanced by new generations of graphics processing chips with optimized “deep learning” neural network capabilities. Many of these accomplishments are making video marketing more sophisticated. Four applications of AI, in particular, are driving some amazing changes in the video field.

1. Enhancing Existing Images and Videos

AI techniques are used to enhance the general quality of still images and videos. They can also be applied to a portion of a scene to repair a missing or damaged area. In the early days of image manipulation, these repairs were often limited to smoothing edges or shading areas, including matching or altering skin tones.

Using today’s sophisticated deep learning algorithms, current software can repair or alter complex human faces in images and videos. After analyzing millions of stored facial images, the software can identify a face in an image; categorize it according to age, ethnicity, or other characteristics; and enhance it to repair or alter specific features such as lips, eyes, nose, or mouth.

A Generative Adversarial Network plays an important role in modern algorithms for creating or reproducing complex, realistic three-dimensional detail. Invented by Google researcher Ian J. Goodfellow, a GAN utilizes a pair of neural networks to learn how to render a complex image such as a human face, and then improve upon that rendering repetitively until it becomes difficult to identify any differences between the original and the newly created image. This technique can be used to replace missing or damaged parts of an image, such as a face within a larger image or an entire image. The relative ease of doing this has raised ethical issues with regard to fake images of all kinds.

2. Creating New Virtual Reality Scenes

Taking AI GAN techniques one step further, graphics and AI chip maker Nvidia is developing software that can take in an entire physical scene and then create a different, realistic scene from it. The software uses deep learning algorithms to analyze a scene from a pre-recorded video or live camera, identifying and learning about objects such as buildings, vehicles, people, and plants. It can then create a completely new scene, given a basic outline of where certain objects should be placed.

This process can be used to create virtual reality environments, including those used to train industrial robots, autonomous vehicles, medical systems, and other computer vision-based devices that interact with the physical world. The technology can also be used to assist with game design, as user demands for more complex and realistic game details show no signs of slowing down. It also has major potential to advance the state of special effects, animation, and graphics in movies.

3. Analyzing Consumer Response to Marketing

AI facial recognition and computer vision techniques can be used for more than altering or filling in facial images. Several marketing firms are using this technology to help their clients test video advertisements to see which are most effective. Big brands including food producers and movie studios are using these services to craft ads that trigger the most positive emotions among their test subjects. The AI algorithms use a camera to monitor changes in a subject’s facial expression as they watch a video ad. In many cases, the software can recognize almost two dozen unique expressions involving specific movements of facial regions, then classifies them as indicating one or more emotions, such as fear, anger, surprise, or joy. This ability helps advertisers determine the best way to target a segment of their audience, as well as revealing which content elicits the strongest reactions.

According to Wyzowl’s State of Video Marketing in 2018 Survey published on HubSpot.com, 81 percent of businesses use video as a marketing tool. 97 percent say video has helped increase user understanding of their product or service, helping 76 percent increase their sales. They also report 81 percent of people have been convinced to buy a product or service by watching a brand’s video. Even in the Business-to-Business marketing space, a recent Contently and Libris survey reports that 33 percent of marketers say that out of all visual content, custom video drives the most engagement and 70 percent drive better marketing ROI when they use visuals. Given the enormous potential for video-based marketing to influence consumers, this application of AI-based computer vision seems likely to continue growing quickly.

4. Extracting Insights From Live Video Search

Going a step further beyond emotion recognition in a controlled test environment, pioneer Youplus utilizes AI machine learning techniques to understand and index video content on the internet, specifically opinions expressed by people in videos. Using an AI-based understanding of human language, emotions, and expressions, the system effectively watches and listens to the videos on the web and collects opinions expressed by people in the videos. Insights into this mountain of data previously buried inside of online videos are then provided by an AI neural network-based search engine called VOISE. Youplus launched this system in 2017. One year later, VOISE is being used by more than 100 marketing and advertising agencies, as well as brands across the United States, India, and the United Kingdom.

Find out more about how VOISE can help you connect more profitably with your customers by visiting Youplus at https://youplus.com.

Sources:

1. https://scholar.google.ca/citations?user=iYN86KEAAAAJ&hl=en
2. https://blog.hubspot.com/marketing/state-of-video-marketing-new-data
3. https://contently.com/resource/libris-visual-content-report/
4. https://youplus.com/voise/