Learning Spatial Relations from Images
Abstract
This paper describes ACORN,1 a system that acquires knowledge about spatial relations by observing raster images. Learned knowledge is represented with constraint programs; this lets the system use its knowledge in a direct and flexible way. After learning, ACORN can either generate scenes from a set of spatial constraints, or find a set of spatial relations that describe a scene. The system works directly with numeric location attributes extracted from a raster image, and transforms these to more approximate symbolic relations.
Cite
Text
Hiraki et al. "Learning Spatial Relations from Images." International Conference on Machine Learning, 1991. doi:10.1016/B978-1-55860-200-7.50084-2Markdown
[Hiraki et al. "Learning Spatial Relations from Images." International Conference on Machine Learning, 1991.](https://mlanthology.org/icml/1991/hiraki1991icml-learning/) doi:10.1016/B978-1-55860-200-7.50084-2BibTeX
@inproceedings{hiraki1991icml-learning,
title = {{Learning Spatial Relations from Images}},
author = {Hiraki, Kazuo and Gennari, John H. and Yamamoto, Yoshinobu and Anzai, Yuichiro},
booktitle = {International Conference on Machine Learning},
year = {1991},
pages = {407-411},
doi = {10.1016/B978-1-55860-200-7.50084-2},
url = {https://mlanthology.org/icml/1991/hiraki1991icml-learning/}
}