Use Computer Vision and Deep Learning to search untagged video
This is a demo of Algorithmia's VideoMetadataExtraction and InceptionNet algorithms. It examines each frame of several videos, autogenerates tags based on their visual content, and makes it possible to search through the videos to find the timepoints at which that tag occurs. For example, a user can search for "chair" and see all videos, and the specific points in time in those videos, where a chair is visible.
See this demo in action
This demo can be viewed at https://demos.algorithmia.com/video-search/
Run it yourself
This demo only contains frontend code and requires no specialized hosting (or even a server).
- download the repository
- if you have not already, run the setup steps outlined in /PUBLISH.md
- build the demo:
- open /build/video-search/index.html in a web browser