AttriGen: Automated Multi-Attribute Annotation for Blood Cell Datasets

2025-09-30

AI Summary

To address the challenges of missing multi-attribute fine-grained annotations, limited model interpretability, and high manual annotation costs in blood cell microscopic imaging, this paper proposes a dual-model collaborative classification framework: a CNN backbone performs cell-type classification, while a Vision Transformer (ViT) models morphological attributes—including size, shape, and texture—enabling decoupled yet joint prediction of cell type and attributes. The framework is co-trained on the PBC and WBCAtt datasets, establishing the first multi-attribute automatic annotation benchmark specifically for blood cells. It achieves a multi-attribute classification accuracy of 94.62%. This approach significantly improves annotation efficiency and model interpretability while reducing reliance on expert labeling, thereby introducing a novel paradigm for fine-grained semantic understanding in medical imaging.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

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.

Problem

• Automates multi-attribute annotation for blood cell datasets
• Enhances classification accuracy and model interpretability in microscopy
• Reduces time and cost compared to human annotation processes

Innovation

• Automated multi-attribute annotation framework for cell microscopy
• Dual-model architecture combining CNN and Vision Transformer
• Achieves high accuracy and enhances model interpretability