Automated Tools for Video Search and Metadata Harvesting

By Patrik - September 25, 2015 (Last updated: August 24, 2017)

An increasing amount of video is produced every day. Most of that video will be seen once, and then forgotten, even though the content may be valuable. Automated tools to find video will be an important part of the future media landscape. This week I read two interesting news on video discovery automation.

First out was news from Valossa, a Finnish start-up, that does deep content discovery in video using audiovisual data and metadata. Together with voice recognition they can answer questions in standard English about movies. Valossa writes on their blog:

“Valossa’s proprietary engine understands both the deep content data of videos and provides novel solutions for content indexing and discovery.”

This ability to automatically index and search in video will increase the value of any large video library, and will be necessary to monetise assets in the future. There are too much video for anyone to manually tag and index already, and it will become even “worse” in the future. Valossa have created a proof-of-concept running on, that you can try out.

The second article this week was from FastCompany, with a write-up of the company Dextro that creates software for automatic discovery of content in video. They have focused content analysis of publicly available videos. From the article:

“To make poorly labeled videos easier to discover, Manhattan-based video analysis startup Dextro is launching a platform that analyzes and tags the contents of publicly available videos”

I love this concept of taking user generated content and make it findable. There are massive amounts of content generated from the general public, from personal videos only to be shared by friends to news-worthy material that should be spread to other’s. It’s hard enough to keep up with the material your friends create, and even harder to get a hold of the rest.

Having technology like this at hand will not only make your content immediately findable (Dextro works in real-time, but will also add valuable metadata to your content, to make it more valuable as  you amass more and more content. We are looking forward to more development in this area, especially when they come with APIs to use for building even more interesting applications.