Image recognition constitutes a key component of machine learning that enables intelligent systems to interpret visual information. Conventional optical character recognition (OCR) tools often encounter challenges with variations in font style, text orientation, language, and the presence of technical symbols. To address these challenges, a convolutional neural network (CNN) was implemented to identify characters individually within annotated labels of electrical circuit images. A post-processing function was integrated to reconstruct the detected characters, ensuring semantic coherence and compliance with standard circuit notation. The proposed method was then compared with a conventional OCR tool to evaluate accuracy and flexibility. Experimental results demonstrate that this approach enhances technical consistency and streamlines the automation of circuit label annotation in intelligent systems.