<![CDATA[The rapid growth of the e-commerce sector has led to a significant surge in shipping volumes in Indonesia. In logistics systems, a shipping receipt serves as a crucial document containing destination information such as address, city/regency, and postal code. Errors or delays in classifying destination regions not only generate additional operational costs (e.g., reshipment fees and service penalties) but may also reduce customer satisfaction and harm the reputation of service providers. This study proposes the implementation of a Convolutional Neural Network (CNN) model to automatically classify region types in shipping receipt images, aiming to minimize manual errors and accelerate processing time. CNN was chosen for its ability to recognize complex visual patterns in digital documents without requiring manual feature extraction. The dataset used in this study consists of 1,540 shipping receipt images from various courier services, labeled as REG_JAWA and REG_LUARJAWA. The research process includes image preprocessing (resizing, normalization, augmentation), CNN architecture design, model training with early stopping, and performance evaluation using accuracy, precision, recall, and F1-score metrics. The results demonstrate that the model achieved a testing accuracy of 99.87%, precision of 99.71%, and recall of 100%, highlighting its strong potential for real-world implementation in logistics systems to improve efficiency and reliability of deliveries.]]>
