Facebook details AI that can understand videos (6 minute read)
Learning from Videos is a project from Facebook that uses AI to automatically learn audio, textual, and visual representations from videos. The project has led to improvements in Facebook's core AI systems and will enable entirely new experiences. This article explores some of Facebook's AI systems, including Generalized Data Transformations, wav2vec 2.0, Audio Visual Textual, TimeSformer, and speech recognition.