Identifying relevant frames in weakly labeled videos for training concept detectors
Adrian Ulges, CHRISTIAN HANSJOACHIM SCHULZE, Daniel Keysers, Thomas Michael Breuel · 2008
A key problem with the automatic detection of semantic concepts (like 'interview' or 'soccer') in video streams is the manual acquisition of adequate training sets. Recently, we have proposed to use online videos downloaded from portals like youtube.com for this purpose, whereas tags provided by users during video upload serve as ground truth annotations.