no code implementations • 30 Jan 2020 • Helard Martinez, M. C. Farias, A. Hines
The approach presented in this work is based on the assumption that autoencoders, fed with descriptive audio and video features, might produce a set of features that is able to describe the complex audio and video interactions.