Real-time Handwritten Digit Recognition Using CNN on Embedded Systems

In the current digital era, the growing demand for efficient digit recognition systems is increasingly evident. The complexity of recognizing handwritten digits, attributed to variations in size, thickness, style, and orientation, now finds solutions through advancements in convolutional neural networks and deep learning. Integrated into embedded systems, this technological progress becomes accessible in everyday life, providing effective solutions for various applications. In this context, this paper aims to implement CNNs in embedded systems, particularly on the Raspberry Pi board, for real-time recognition of handwritten digits. We will closely examine the architecture, design software, and datasets used in this field. Finally, we will address the implementation of CNNs for instant digit recognition, concluding with the successful identification of multiple digits.