A team from Disney Research and Shanghai's Fudan University has used deep-learning techniques to train computer software to recognize events in videos. They say the techniques can enable software to recognize categories of events that it has not previously seen. The approach, which uses both scene and object features from the video, enables associations to be made between these visual elements and each type of event to be automatically determined and weighted by a neural network. "The model learns what is important for each event," says Disney Research postdoctoral researcher Yanwei Fu.
When presented with an event that it has not previously encountered, the model can identify objects and scenes that it already has associated with similar events to help it classify the new event, says Disney Research's Leonid Sigal. He also notes the technique works better than other methods in recognizing events in videos. Sigal says the "events" can include riding a horse, baking cookies, or eating at a restaurant.
"Automated techniques are essential for indexing, searching, and analyzing the incredible amount of video being created and uploaded daily to the Internet," says Disney Research's Jessica Hodgins. "And if we don't know what's in all those videos, we can't find things we need and much of the videos' potential value is lost."
From Disney Research
View Full Article
Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA