The average knowledge worker consumes hours of digital media a day in the form of blogs, television broadcasts, Internet videos and more.
Digital media is certainly easy to consume but, according to Brad Sousa, Chief Technology Officer at AVI Systems, it’s not very easy to sort, tag and file in a way that makes it easy to find and monetize later. At least, not for people. But video data mining is primed to be the next frontier of artificial intelligence.
Using AI To Capture and Use Video Data
Video data projects that incorporate AI, natural language processing and machine learning are largely still in the research and testing phase. But they show a lot of potential for increasing searchability and usability, as well as for targeting digital media to consumers in real-time.
Check out these up-and-coming AI applications to video data mining and application.
Personalization through video recognition
Many consumers are familiar with voice recognition thanks to natural language processing applications such as Siri, Alexa and Google Home. These applications can recognize your voice and respond to voice commands to perform tasks and set systems, say a conference room AV system, to a specified user preference. Video recognition capabilities paired with AI take it to the next level.
Instead of using your voice to activate a conference room AV system after you are in the room, what if an AI-enabled camera at the entrance of the conference room captured an image of your face as you entered, identified you and had the system up and running based on your preferences before you even sat down?
Now, let’s take it a step further. What if that AI-enabled facial recognition capability was connected to the conference room scheduling system? Now the AI knows who scheduled the meeting and can differentiate that person from other participants to start and control the meeting room system accordingly.
Improved searchability through AI meta tagging
Organizations in industries such as national security, education, and health care have an enormous volume of visual and digital media, training videos, lectures or security footage. The issue is, how do you tag and sort that much data in a way that makes it efficient and effective to search and use again?
It would take a CPU nearly two hours to watch and tag a 60-minute video, but it takes an AI engine just 10 minutes to annotate and index the same thing. Sophisticated meta tagging of digital media makes it easily searchable for just about anyone. A college student taking classes remotely could quickly search the school’s entire digital archive for a recorded lecture, or a health care provider could quickly access an instructional video to walk them through an emergency procedure in real-time.
Targeted media offerings for consumers and advertisers
Companies and, in particular, advertisers can learn a lot about consumers based on the media they search for. But they could learn even more about users, and better monetize the media they consume if they knew how users reacted to that media in real-time.
If broadcasters and other video content providers could see that a viewer’s attention was fading or could observe the reaction a viewer has when a certain actor or product was on the screen, they could adjust content in real-time, provide more targeted advertising opportunities and increase monetization of online and streaming content.
Keeping AI Ethical—And Comfortable
Perhaps the greatest challenge faced by AI is walking the line between improving the consumer experience and creeping consumers out, Sousa said.
While Millennials have a low expectation of privacy and anonymity both on- and offline, Baby Boomers and workers from previous generations might not like the idea of a camera recognizing their face or an AV system knowing personal preferences like how they want a meeting set up.
Currently, if a person has voluntarily provided information and access to their location, preferences and other personal data, the related AI and data mining capabilities are on the “right” side of the line. The ongoing ethics discussion should be part of any AI-enabled technology development or implementation, Sousa said.
A qualified AV integrator can talk you through all the implications of AI capabilities and help you choose the right—and safest—solution.