120
click to vote
DAGM
2008
Springer
15 years 4 months ago
2008
Springer
Abstract. In this paper, we propose a new approach to learn structured visual compound models from shape-based feature descriptions. We use captioned text in order to drive the pro...