Just a few months after unveiling its software to analyze live video in real-time, Dextro is trying to up its game in the somewhat frothy video discovery sector.
CEO and co-founder David Luan says his New York-based company took the wraps off a new service Thursday that helps businesses with extensive video libraries make their content discoverable. “We are now launching Sight, Sound and Motion, which is a unified model of everything that’s happening in the video,” Luan says.
Previously, searches were done through computer vision algorithms, he says, that only analyzed the visuals of videos to determine their contents. Luan says his company realized the audio track and motion cues in videos could also be used to sort out what is happening onscreen and make the content more discoverable. That means the motion of a touchdown pass being caught in a football game or the name of a celebrity being spoken can be used to find videos.
Sifting through the ever-growing mountain of user-generated content (UGC) streamed and shared on the Web can be daunting, to say the least. Despite the renewed attention brought to the sector by live-streaming apps Periscope and Meerkat, there have been a few red flags raised about the video discovery scene.
A couple of years ago, Hunter Walk, the former director of product management at YouTube, called out what he believed were reasons why video discovery startups fail. Issues with scalability topped his list, but he went on to cite a need for context with videos rather than simply creating collections. He also pointed out the challenge of trying to monetize third-party content that cannot be directly monetized.
Back in February, San Francisco-based N3twork pivoted its focus from its video social network to mobile gaming. Others, such as StumbleUpon’s 5by, keep chugging along, though—with a tweak or two.
Last October, 5by relaunched and expanded from being a video curation app that matched content with the users’ moods, to one that also encouraged sharing and chatting about the videos with friends.
Dextro has taken a different approach, offering its services to companies like media outlet Mic.com, Luan says, which use it to power searches of content. Mic.com draws upon citizen journalists to capture in-person footage of current events, such as the refugee crisis in Croatia.
Luan says the Sight, Sound and Motion service helps them comb through that content, and offers more context than the sparse, text-based metadata typically used to identify and search for videos. “It’s a way to distill the mass of UGC down to what they just want to see,” Luan says. “That puts us closer to being the backend infrastructure for Web-video search for our customers.”
