Shutterstock launches reverse image search for video
Stock library Shutterstock has launched a new Reverse Image Search feature for video, which should make identifying clips easier and faster.
Designed to be more responsive than traditional keyword searches, the software uses machine learning to reference an archive of more than 280 million images and more than 15 million video clips at the individual pixel level.
Users can simply upload an image of their choosing into the drag-and-drop style interface, and once the software has recognised what’s inside each image, it visually searches the database (from the submitted image) to find similar clips that match the image's basic composition, content and tone.
To test it out, we tried uploading a fairly generic cycling image featuring a very distinctive rock wall. The results were pretty impressive, with the reverse video search clearly identifying clips with similar tonal values to the wall, and although it didn't pick up the fact there was a cyclist in our frame, it did identify there was a human figure.
For video editors looking for stock footage to complete a project, the new software should be a big timesaver. Shutterstock's Reverse video search feature is available now. You can try it for yourself at Shutterstock.