Handwritten digit recognition based on channel and spatial attention mechanisms
Abstract
Since handwritten digits are not as clear and standardized as printed characters, enabling computers to recognize them poses certain challenges. by adopting a channel and a spatial attention mechanism a deep learning networks is designed to achieve training on the MNIST dataset and recognition of handwritten digit images. The verified experiments demonstrate that the designed deep learning network system achieves high efficiency in data processing, global optimization capability, and strong generalization performance.
Fields of Study & Subjects
Handwritten Text Recognition Techniques
Advanced Neural Network Applications
Vehicle License Plate Recognition
Publication details
- Journal / source
- International Symposium on Robotics and Artificial Intelligence (ISRAI)
- Pages
- 53–53
- Publication date
- 12 February 2026
- Language
- EN