Lot information plays an essential role in tracking product information. Errors in printing lot information (such as lot numbers, manufacturing dates, and expiration dates) on product packaging often result from inaccuracies in the printing line, printer quality, or human error. These issues can negatively affect a company’s brand if such defective products are released to the market. In this paper, an integrated automatic classification system is proposed to classify defective and non-defective printed lot numbers on the existing lot printing line in a factory. This study focuses on the use of optical character recognition (OCR) models, combined with image preprocessing, model training, and fine-tuning, to improve the accuracy of recognizing printed information. The test dataset was collected from lot information printed on real product packaging in the factory. Experimental results demonstrate that the proposed method is effective in recognizing lot information on product packaging. Furthermore, by applying classification and comparing the OCR results with the original reference lot information, the system can efficiently detect defective samples with a noticeable false negative rate on the test set. It is expected that the proposed approach will provide a low-cost and practical inspection solution that can be easily deployed in real-world manufacturing environments.