AttriGen: Automated Multi-Attribute Annotation for Blood Cell Datasets

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Houmaidi, Walid, Sabiri, Youssef, Iguenfer, Fatima Zahra, Abouaomar, Amine
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866912618030039040
author Houmaidi, Walid
Sabiri, Youssef
Iguenfer, Fatima Zahra
Abouaomar, Amine
author_facet Houmaidi, Walid
Sabiri, Youssef
Iguenfer, Fatima Zahra
Abouaomar, Amine
contents We introduce AttriGen, a novel framework for automated, fine-grained multi-attribute annotation in computer vision, with a particular focus on cell microscopy where multi-attribute classification remains underrepresented compared to traditional cell type categorization. Using two complementary datasets: the Peripheral Blood Cell (PBC) dataset containing eight distinct cell types and the WBC Attribute Dataset (WBCAtt) that contains their corresponding 11 morphological attributes, we propose a dual-model architecture that combines a CNN for cell type classification, as well as a Vision Transformer (ViT) for multi-attribute classification achieving a new benchmark of 94.62\% accuracy. Our experiments demonstrate that AttriGen significantly enhances model interpretability and offers substantial time and cost efficiency relative to conventional full-scale human annotation. Thus, our framework establishes a new paradigm that can be extended to other computer vision classification tasks by effectively automating the expansion of multi-attribute labels.
format Preprint
id arxiv_https___arxiv_org_abs_2509_26185
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AttriGen: Automated Multi-Attribute Annotation for Blood Cell Datasets
Houmaidi, Walid
Sabiri, Youssef
Iguenfer, Fatima Zahra
Abouaomar, Amine
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
60G35, 62M10, 62P35, 65C20, 68T45, 68U10, 92C35, 92C40, 92C42, 93E10
I.4; I.4.8; I.4.9; I.4.10; I.2; I.2.6; I.2.10; J.3; C.2.4; C.3; H.2.8; H.3.4; H.3.5; I.2.4; I.5; I.5.1; I.5.4; K.6.1
We introduce AttriGen, a novel framework for automated, fine-grained multi-attribute annotation in computer vision, with a particular focus on cell microscopy where multi-attribute classification remains underrepresented compared to traditional cell type categorization. Using two complementary datasets: the Peripheral Blood Cell (PBC) dataset containing eight distinct cell types and the WBC Attribute Dataset (WBCAtt) that contains their corresponding 11 morphological attributes, we propose a dual-model architecture that combines a CNN for cell type classification, as well as a Vision Transformer (ViT) for multi-attribute classification achieving a new benchmark of 94.62\% accuracy. Our experiments demonstrate that AttriGen significantly enhances model interpretability and offers substantial time and cost efficiency relative to conventional full-scale human annotation. Thus, our framework establishes a new paradigm that can be extended to other computer vision classification tasks by effectively automating the expansion of multi-attribute labels.
title AttriGen: Automated Multi-Attribute Annotation for Blood Cell Datasets
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
60G35, 62M10, 62P35, 65C20, 68T45, 68U10, 92C35, 92C40, 92C42, 93E10
I.4; I.4.8; I.4.9; I.4.10; I.2; I.2.6; I.2.10; J.3; C.2.4; C.3; H.2.8; H.3.4; H.3.5; I.2.4; I.5; I.5.1; I.5.4; K.6.1
url https://arxiv.org/abs/2509.26185