Non-Label-Based Goods Identification in Large-Scale Warehousing and Automated Logistics Operations Using Vision-Based OCR

Background: Automated identification of logistics units is a critical requirement in high-volume warehouse operations, particularly in retails that handle millions of cartons annually. Although barcode-based systems are widely used, they generate recurring costs for labeling, printing, quality control and readability issues, often leading to manual intervention and delays. Methods: This study presents a low-cost and flexible vision-based identification system that directly reads carton identifiers using optical character recognition (OCR). This system designed for edge deployment on resource-constrained hardware and incorporates a rotation-invariant preprocessing pipeline to support robust recognition under real conditions. Proposed approach was tested in two German retails. Results: Tests show recognition accuracies 96% to 98% under operational conditions, with real-time processing performance in the range of 58 to 125 ms per scan, depending on the hardware. These indicate that the system can be integrated into high-throughput logistics workflows. Additionally, the study provides insights into the economic implications of replacing barcode-based identification. Based on site-specific observations and labeling costs, the system shows the potential to reduce manual intervention and lower operational expenses in large-scale retails. Conclusions: Findings suggest that OCR can serve a cost-efficient alternative to barcode systems in environments where flexibility, robustness, and low deployment cost are critical.