Unsupervised Linking of Visual Features to Textual Descriptions in Long Manipulation Activities.

Semantic Decomposition and Recognition of Long and Complex Manipulation Action Sequences.