Video Content can be stored on client servers or uploaded directly to Fovea. Content is processed using state-of-the-art machine learning algorithms to extract metadata such as people, text, and objects.
Users can enter a text query, describing their desired video. Our AI-based system automatically assesses what the text is about and selects clips from an existing database to cut together a video.
Users can upload just a single style reference video, and our system intelligently learns the style features of that video, allowing it to edit future videos with similar visual coherency.
Content is delivered dynamically. For example, the video player can change with new incoming footage, or edits can be targeted to audiences based on their location.
We founded Fractal Media in NYC, a creative video agency that provides client advertising services and produces original digital content. While running Fractal, we developed a number of internal tools to automate various aspects of the video production process.
Over the course of the summer of 2019, we developed the first prototype of Fovea, which utilized eye tracking glasses provided by Pupil Labs. This began with developing a machine learning model to process eye tracking data and was followed by building of a custom video player for content playback based on the outputs of that model.
In the fall of 2019, we presented Fovea at the NYC Media Lab Summit Demo Expo, a 'science fair' style expo of over 100 demos presented to industry and media figures. And we won second prize overall!
In January 2020, we joined the NYC Media Lab's Combine Accelerator program.
In summer 2020 we developed our automated video editing system, and are currently planning on launching Fovea in the comping months.
Eye Tracking for Interactive Video Presentation
We’re sharing updates as we head towards our launch and in this post, we present an overview of our platform, and describe the technology used on the research side.
Interactive, Remote Training
If online education and remote training in 2020 has proven anything, learning management systems online have mostly failed to match up to in-person classes and seminars. Good teachers provide interactive learning, make students feel as if their attention matters and are able to design and aggregate educational material intelligently.
Content Delivery Over-the-Web Using Eye Tracking
By combining tools in object detection and eye tracking, we will aim to queue video, text and images containing people, objects and imagery to keep engagement high. This type of real-time personalization is the future of media delivery, and has immediate, practical applications as the amount of news content online grows.
Founder
Sharath Koorathota is the co-founder and software engineer at Fovea. He is passionate about computer science, and has spent the last 5 years understanding how artificial intelligence can provide personalized content.
sharath@foveainsights.com
Founder
Patrick Adelman is a founder and software engineer at Fovea. Over the past 7 years Patrick has worked in advertising, in roles as a producer, editor and motion graphics artist. He has worked on projects for digital, social, and television.
patrick@foveainsights.com26 Broadway, 8th FloorNew York, NY 10004info@foveainsights.com