Ryota Tanaka
2024
How Well Do Vision Models Encode Diagram Attributes?
Haruto Yoshida
|
Keito Kudo
|
Yoichi Aoki
|
Ryota Tanaka
|
Itsumi Saito
|
Keisuke Sakaguchi
|
Kentaro Inui
Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop)
Research on understanding and generating diagrams has used vision models such as CLIP. However, it remains unclear whether these models accurately identify diagram attributes, such as node colors and shapes, along with edge colors and connection patterns. This study evaluates how well vision models recognize the diagram attributes by probing the model and retrieving diagrams using text queries. Experimental results showed that while vision models can recognize differences in node colors, shapes, and edge colors, they struggle to identify differences in edge connection patterns that play a pivotal role in the semantics of diagrams. Moreover, we revealed inadequate alignment between diagram attributes and language representations in the embedding space.
Search
Co-authors
- Haruto Yoshida 1
- Keito Kudo 1
- Yoichi Aoki 1
- Itsumi Saito 1
- Keisuke Sakaguchi 1
- show all...
Venues
- acl1