Research on an Algorithm of Express Parcel Sorting Based on Deeper Learning and Multi-Information Recognition.

Area of Science:

• Logistics and Supply Chain Management
• Computer Vision
• Artificial Intelligence

Background:

• Smart logistics centers face challenges with low-cost sorting equipment and complex warehouse environments.
• Current single-information identification methods, like basic barcode reading, lead to reduced sorting accuracy and efficiency.
• The need for robust and accurate express recognition in automated logistics is critical.

Purpose of the Study:

• To propose a novel express recognition method integrating deep learning and multi-information fusion.
• To enhance the accuracy and efficiency of express sorting in smart logistics environments.
• To address the limitations of single-source data identification in complex warehouse settings.

Main Methods:

• Developed a deep learning-based target detection network, improving upon the YOLOv4 architecture for speed and precision.
• Implemented a multi-information fusion approach combining barcode data and three-segment code recognition.
• Utilized the ZBAR algorithm for barcode decoding and Tesseract-OCR for three-segment code identification.

Main Results:

• Achieved a detection time of 0.31 seconds per image.
• Reached a high recognition accuracy of 98.5%.
• Demonstrated superior robustness and accuracy compared to single barcode recognition methods.

Conclusions:

• The proposed deep learning-based multi-information identification method significantly improves express sorting accuracy and efficiency.
• This approach offers a more robust and accurate solution for logistics centers compared to traditional methods.
• The enhanced recognition system aids in accurately obtaining express sorting information from databases.