Guardado en:
Detalles Bibliográficos
Autores principales: Beňová, Ivana, Gregor, Michal, Gatt, Albert
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2409.01389
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912480054214656
author Beňová, Ivana
Gregor, Michal
Gatt, Albert
author_facet Beňová, Ivana
Gregor, Michal
Gatt, Albert
contents How do vision-language (VL) transformer models ground verb phrases and do they integrate contextual and world knowledge in this process? We introduce the CV-Probes dataset, containing image-caption pairs involving verb phrases that require both social knowledge and visual context to interpret (e.g., "beg"), as well as pairs involving verb phrases that can be grounded based on information directly available in the image (e.g., "sit"). We show that VL models struggle to ground VPs that are strongly context-dependent. Further analysis using explainable AI techniques shows that such models may not pay sufficient attention to the verb token in the captions. Our results suggest a need for improved methodologies in VL model training and evaluation. The code and dataset will be available https://github.com/ivana-13/CV-Probes.
format Preprint
id arxiv_https___arxiv_org_abs_2409_01389
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle CV-Probes: Studying the interplay of lexical and world knowledge in visually grounded verb understanding
Beňová, Ivana
Gregor, Michal
Gatt, Albert
Computation and Language
How do vision-language (VL) transformer models ground verb phrases and do they integrate contextual and world knowledge in this process? We introduce the CV-Probes dataset, containing image-caption pairs involving verb phrases that require both social knowledge and visual context to interpret (e.g., "beg"), as well as pairs involving verb phrases that can be grounded based on information directly available in the image (e.g., "sit"). We show that VL models struggle to ground VPs that are strongly context-dependent. Further analysis using explainable AI techniques shows that such models may not pay sufficient attention to the verb token in the captions. Our results suggest a need for improved methodologies in VL model training and evaluation. The code and dataset will be available https://github.com/ivana-13/CV-Probes.
title CV-Probes: Studying the interplay of lexical and world knowledge in visually grounded verb understanding
topic Computation and Language
url https://arxiv.org/abs/2409.01389