algorithms - PARC-projects/video-query-home GitHub Wiki
Video Query (Client, API and Algorithms) enables agile, user-in-the-loop mining of video datasets. Rather than requiring an upfront specification of actions and investment in labeling data for those actions, Video Query embraces early exploration of evolving questions and ideas. The effort spent increases only as the user iterates to, focuses down on, and validates a particular line of inquiry. Video Query enables agile browsing through a large dataset, in order to discover and zoom in on relevant data in a dynamic and interactive way.
The Video Query Algorithms repository provides algorithms and code for
- building a database of embedded CNN features from videos,
- brokering search requests from the Video Query application,
- searching the database for video clips similar to a provided reference, and
- updating the weighted criteria for selecting matches.
Video Query API uses Video Query Algorithms to search for and propose matching video clips. Video Query Client is a frontend that uses endpoints at Video Query API. Video Query API can also be used independently of Video Query Client. The broker in Video Query Algorithm polls the API for the process state of queries, and acts on those queries which are waiting for a new or revised set of matches.