Transfer Learning for Russian Handwriting Recognition

Abstract

Handwriting recognition is one of the most important tasks in computer vision. Despite the variety of methods and algorithms used for text recognition in various languages, the recognition of handwritten Russian words is still a completely unsolved problem, which keeps the relevance of this topic. This paper proposes the application of a convolutional neural network (CNN) to the task of recognizing the 100 most frequently used Russian words, trained on a new unique dataset. A unique algorithm for extracting Russian handwritten words from forms was applied to obtain a new dataset. The paper also presents the results of a comparison of such powerful models as VGG-16, VGG-19, ResNet-50, and ResNet-101, which were trained using transfer learning technology. The best results were shown by VGG-16 with 98.7% accuracy, and the worst results were shown by ResNet-50 with 80% accuracy.

More by Dmitry Kaplun

• Codewithsayanjib/slm-systematic-review: v1.1Zenodo · 2026
• Microscopy Image Dataset of pulmonary vessels for Quantitative assessment of fibrosisScientific Data · 2026
• Microscopy Image Dataset of pulmonary vessels for Deep Learning-Based Quantitative assessment of fibrosisFigshare · 2026
• MA2MU-Net++: A Mahalanobis distance and Ackley function based multi-modal U-Net++ model for breast nuclei segmentation in histopathology imagesBiomedical Signal Processing and Control · 2026