The speaker-gestures-based information retrieval for lecture
videos has been studied in [16]. The author equipped
the lecture speaker with special gloves that enable the automatic
detection and evaluation of gestures. The experimental
results show that 12 percent of the lecture topic
boundaries were correctly detected using speaker-gestures.
However, those gesture features are highly dependent on
the characteristics of speakers and topics. It might have limited
use in large lecture video archives with massive
amounts of speakers.