Quality control in manufacturing logistics with convolutional neural networks

Digital transformation revolutionizes business processes, manufacturing, and supply chain management. It creates new business models, making the contemporary market more competitive and dynamic. Companies that follow the trend of digitalization gain additional competitive advantages due to a quick responses to consumer needs and market changes. Quality control in manufacturing logistics is one of the business processes that may benefit considerably from applying digital solutions. However, quality control of packed finished goods in manufacturing logistics remains a relatively less explored area, despite its importance for real-world industrial applications. Existing studies in computer vision and logistics typically address related tasks such as object detection, barcode recognition, or general defect detection, but only partially address the specific challenges associated with inspecting packaged goods in complex industrial environments. The applied value of the present research lies in the developed algorithms for defect detection and classification. The suggested approach combines deep learning techniques with economic efficiency evaluation methods. This approach covers technical implementation and business-level return on investment. The empirical contribution of the study considers the actual needs of a manufacturing enterprise and includes dataset preparation, model-development results, and an implementation framework. The findings indicate the technical feasibility of the proposed framework, offering valuable experience for similar contexts.