We are investigating how the power of crowd-workers can be leveraged to validate, augment and partition metadata on large quantities of video. Using the Bootlegger platform as a data-store and initial set of context based meta data, we are developing unit task designs that deliver efficient, rich and high quality metadata at a sub-clip level.
- Focus and audio checking
- Correction of shot type
- Tracking individuals through footage
- Subjecting content testing (crowd editing)
- Audio description
- Crowdflower API