Learning complete and explainable visual representations from itemized text supervision

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Lyu, Yiwei, Zhao, Chenhui, Banerjee, Soumyanil, Liu, Shixuan, Rao, Akshay, Kondepudi, Akhil, Lee, Honglak, Hollon, Todd C.
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866911520663797760
author Lyu, Yiwei
Zhao, Chenhui
Banerjee, Soumyanil
Liu, Shixuan
Rao, Akshay
Kondepudi, Akhil
Lee, Honglak
Hollon, Todd C.
author_facet Lyu, Yiwei
Zhao, Chenhui
Banerjee, Soumyanil
Liu, Shixuan
Rao, Akshay
Kondepudi, Akhil
Lee, Honglak
Hollon, Todd C.
contents Training vision models with language supervision enables general and transferable representations. However, many visual domains, especially non-object-centric domains such as medical imaging and remote sensing, contain itemized text annotations: multiple text items describing distinct and semantically independent findings within a single image. Such supervision differs from standard multi-caption supervision, where captions are redundant or highly overlapping. Here, we introduce ItemizedCLIP, a framework for learning complete and explainable visual representations from itemized text supervision. ItemizedCLIP employs a cross-attention module to produce text item-conditioned visual embeddings and a set of tailored objectives that jointly enforce item independence (distinct regions for distinct items) and representation completeness (coverage of all items). Across four domains with naturally itemized text supervision (brain MRI, head CT, chest CT, remote sensing) and one additional synthetically itemized dataset, ItemizedCLIP achieves substantial improvements in zero-shot performance and fine-grained interpretability over baselines. The resulting ItemizedCLIP representations are semantically grounded, item-differentiable, complete, and visually interpretable. Our code is available at https://github.com/MLNeurosurg/ItemizedCLIP.
format Preprint
id arxiv_https___arxiv_org_abs_2512_11141
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning complete and explainable visual representations from itemized text supervision
Lyu, Yiwei
Zhao, Chenhui
Banerjee, Soumyanil
Liu, Shixuan
Rao, Akshay
Kondepudi, Akhil
Lee, Honglak
Hollon, Todd C.
Computer Vision and Pattern Recognition
Training vision models with language supervision enables general and transferable representations. However, many visual domains, especially non-object-centric domains such as medical imaging and remote sensing, contain itemized text annotations: multiple text items describing distinct and semantically independent findings within a single image. Such supervision differs from standard multi-caption supervision, where captions are redundant or highly overlapping. Here, we introduce ItemizedCLIP, a framework for learning complete and explainable visual representations from itemized text supervision. ItemizedCLIP employs a cross-attention module to produce text item-conditioned visual embeddings and a set of tailored objectives that jointly enforce item independence (distinct regions for distinct items) and representation completeness (coverage of all items). Across four domains with naturally itemized text supervision (brain MRI, head CT, chest CT, remote sensing) and one additional synthetically itemized dataset, ItemizedCLIP achieves substantial improvements in zero-shot performance and fine-grained interpretability over baselines. The resulting ItemizedCLIP representations are semantically grounded, item-differentiable, complete, and visually interpretable. Our code is available at https://github.com/MLNeurosurg/ItemizedCLIP.
title Learning complete and explainable visual representations from itemized text supervision
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.11141