Joe Ellis and Dan Morozoff have been fascinated with the way people obtain information from videos and how these videos lead to our understanding of the world. The pair met four years ago as doctoral students at Columbia University on a project called News Rover. Along with their advisor, Shih-Fu Chang, the pair and colleagues set out to answer the question of why viewership in television news has drastically declined over the past 10 years, and how they could solve this problem. They decided that the internet has fundamentally changed the way people consume news. In today’s society people only want news they are interested in, on-demand, and from perspectives they trust. This hardly describes the paradigm that currently exists in television news.
To solve this problem, Dan and Joe built a television news processing system called News Rover, that ingested up to 100 hours of raw TV news a day and then organized video clips based on information within them. For example, the system could organize the clips based on what news event was being discussed, who was actually speaking, what they were saying, and what was appearing on screen. News Rover won multiple academic and industry accolades including 1st Place in the ACM Multimedia Conference Grand Challenge in Barcelona. The technology developed was patented by Columbia University, and the team started thinking about possible commercial applications of their academic achievements. Through the News Rover project, the team have published and patented foundational research in machine learning, computer vision, multimodal information processing, and multimedia.
Dan and Joe joined the NYC Media Lab’s Combine Incubator program in January to try and understand whether the News Rover technologies could be transformed into a viable business. They learned that content creators are severely under monetizing this portion of their business, because the internal video content management solutions for searching, recommending, and disseminating their video content are currently inadequate. To address these needs, Dan and Joe have founded Vidrovr to bring the technology they developed to the people who need it most.
Vidrovr can index, search, and recommend video content in a cost-effective, automatic, and accurate manner. Vidrovr addresses three key market needs: 1. Domain and customer-specific automatic metadata generation for videos, 2. Video Content Management solutions that enable automatic placement and recommendation of video clips for across a company’s digital products, and 3. Automatically linking and sourcing visual social media content that is relevant to a particular video or online article before it is published. Vidrovr was recently named as one of the winners of the prestigious Publicis90 competition, which entails investment and mentorship from Publicis Groupe.