Year: 2025
Type: dissertation
Abstract: The paper examines the task of automatic classification of leukocytes in blood smears using modern deep learning architectures. The novelty of the research lies in the comparative analysis of the effectiveness of three models: EfficientNet (a convolutional neural network), Vision Transformer (ViT), and Swin Transformer (transformer architectures) for this medical task. The aim of the work is to determine the most accurate and efficient model for automated blood analysis. To achieve this goal, the following tasks were completed: preprocessing of blood smear images, training and validation of the models, and comparison of their metrics (accuracy, F1-score). The methods used include transfer learning, data augmentation, and cross-validation.
Authors: Гордон, А. Д., Gordon, A. D.
Language: Russian
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7115035875