GLS
Logistics
Computer Vision System for Information Retrieval from Package Labels
In the logistics sector, efficiency in handling and distributing packages is crucial to maintaining top-tier service and ensuring customer satisfaction, as any delay or error can significantly impact the supply chain and the company’s reputation.
Shipping labels play a fundamental role in package management, containing vital information for the correct delivery of shipments. However, it is common for some labels to be unreadable due to damage, poor printing, or wear during transportation and handling, which can cause delays and errors in delivery.
GLS, a leading company in logistics solutions, identified the need to improve their label reading systems to ensure more efficient and accurate package management, even when labels are not legible.
Challenge
GLS faced the challenge of developing a tool that could retrieve information from damaged or poorly printed shipping labels. The goal was to create a computer vision system, integrated into their megatronics label reading machine, that could read and process images of these labels, extract the necessary information, and categorize it correctly for subsequent use in less than 300 milliseconds. Additionally, it was crucial for this system to integrate effectively with GLS’s existing systems, providing a robust and reliable solution that could be deployed locally.
The first step involved receiving and analyzing real examples of shipping labels, developing an engine capable of preprocessing these images, recognizing and interpreting the characters, and deploying this engine in an API for testing. This development needed to be completed within six weeks from the receipt of the required information.
Solution
Foqum developed an innovative solution for GLS by adapting its IDP/OCR algorithms, creating a computer vision system for reading and processing damaged or poorly printed shipping labels for GLS.
This system includes a preprocessing engine that cleans and prepares the label images, followed by an advanced character recognition system that extracts and categorizes the information. The engine is deployed via a REST API, allowing for integration and testing on GLS’s local server.
The project development included a preprocessing engine and a character recognition system. Upon deployment, the engine was piloted on GLS’s local server, accompanied by functional documentation and detailed metrics.
Thanks to this solution, GLS can efficiently handle packages with damaged labels, ensuring that crucial information is retrieved and used appropriately for accurate delivery in record time, increasing productivity per package line. This improvement enhances GLS’s operational efficiency, reduces delivery errors and delays, and improves customer satisfaction and the company’s reputation in the competitive logistics sector.
