Object and scene recognition software work together to understand video content

Researchers from Disney Research and Shanghai’s Fudan University have used deep learning techniques to train computer software to recognize events in videos, even categories of events that the software has not previously seen. Their approach uses both scene and object features from the video and enables associations between these visual elements and each type of event to be automatically determined and weighted by a machine-learning architecture known as a neural network. Notably, this approach not only works better than other methods in recognizing events in videos, but is significantly better at identifying events that that the computer program has never or rarely encountered previously, said Leonid Sigal, senior research scientist at Disney Research. These events can include such things as riding a horse, baking cookies or eating at a restaurant.…


Link to Full Article: Object and scene recognition software work together to understand video content