Meet French startup Reminiz, a computer vision company that can index any type of video — it’s a sort of Googlebot, but for video content. Reminiz can add tags of people, logos or emotions on live streams and on-demand videos.
“The web is designed so that you can search for text — not video. We are making it possible to search within videos,” co-founder and CEO Jack Habra told me.
There are a few different use cases for Reminiz. First, the company works with broadcasters and telecom companies. For instance, Reminiz has a partnership with Orange so that you can learn more about who’s on the screen right now. It could potentially be leveraged for recommendations or contextual ads for external content.
Reminiz streams live channels on its servers directly, scans images and adds tags. Users then download metadata from the servers.
Second, you can use Reminiz to promote your brand on relevant videos. For instance, Hyundai sponsors Lyon’s soccer team. It wants to distribute Hyundai ads before soccer footage with the team playing. But YouTube keywords aren’t that good when it comes to targeting such a specific audience — a video might talk about the soccer team without showing any actual footage.
Brands can then whitelist videos to distribute ads on those videos in particular. You get charged based on minutes of video footage processed by Reminiz.
The company competes with AWS Rekognition and other generic video analysis APIs from cloud providers. What makes Reminiz stand out is that the company builds its own database of faces, people, brands and tags. It’s also probably easier to implement Reminiz compared to a more generic solution.
“With GDPR, everybody is contacting us to focus more on contextual data instead of personal data,” Habra said.