In the current era of manufacturing digitalization, computer vision is an emerging field of artificial intelligence (AI) that replicates human visual system functions by enabling computers and systems to identify and interpret visual information. Convolutional Neural Networks (CNNs) have demonstrated remarkable potential in image recognition and tasks related to pixel data processing. Industrial parts recognition and classification systems play a crucial role in automating manufacturing processes, streamlining workflow, and enhancing object recognition capabilities on production lines. This research presents a CNN-based approach to industrial parts recognition and classification, offering new possibilities for efficient and accurate prediction performance. The Inception-v3 network architecture was selected for its state-of-the-art design, serving as the pre-trained model to achieve optimal performance. The TensorFlow library provided a comprehensive and flexible framework, while dropouts were employed to prevent overfitting. The trained CNN model was evaluated using key metrics such as accuracy, precision, recall, F1-score, and model loss, with a confusion matrix adopted for binary classification to assist decision-making and summarize the model’s predictions. The study results demonstrated that the model achieved an impressive accuracy of 98.83% with a model loss of 0.023084, proving highly effective in identifying industrial parts, specifically distinguishing between BJ15–860 and BJ15–870 screws. This research underscores the capabilities of artificial intelligence, particularly in CNNs, and highlights the application of industrial parts recognition and classification, contributing significantly to the digitalization of manufacturing processes.