A Smart Package Pickup Method and System for Postal Stations Based on Image Recognition
By combining image recognition and large language models, the intelligent parcel pickup system at the station has achieved automated positioning and visual guidance, solving the problems of congestion and mis-pickup during peak hours, and improving pickup efficiency and user experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI YUANQING INFORMATION TECH CO LTD
- Filing Date
- 2026-01-13
- Publication Date
- 2026-06-02
AI Technical Summary
Existing parcel pickup stations suffer from problems such as congestion during peak hours, long pickup times, and high rates of misdelivery. In particular, with high parcel volumes, it is difficult to achieve automated and intelligent parcel location and retrieval.
Image recognition technology is used to detect the express delivery label area through the YOLOv8s model, text extraction is performed by EasyOCR, and semantic understanding is performed by the Vllm+QWen3-8B large language model to achieve automatic matching and visual guidance of express codes. Service interfaces are provided through LangChain and FastAPI to support smart pickup requests from mobile devices.
It improved package pickup efficiency, shortened user waiting time, reduced the rate of mis-pickup, enhanced system scalability, and achieved full automation and intelligence of the package pickup process at the station.
Smart Images
Figure CN122135276A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of express logistics service technology, and in particular relates to a smart parcel pickup method and system for express stations based on image recognition. Background Technology
[0002] With the rapid development of e-commerce logistics, last-mile delivery service stations (such as Cainiao Station, Mama Station, and Fengchao) have become an indispensable part of residents' daily lives. However, in actual use, "difficulty in finding packages" has become a prominent problem commonly reported by users, seriously affecting user experience and station operational efficiency. Stations handle a large number of packages, averaging thousands per day during peak periods, but lack scientific classification: concentrated pickup times during get off work and school hours lead to queues and congestion, making the process of finding packages time-consuming and further extending waiting times, causing user dissatisfaction. Furthermore, some stations have cramped spaces, posing safety hazards when crowded. Due to similar package appearances or unclear labels, users can easily mistakenly take someone else's package, all of which greatly affect pickup efficiency and cause inconvenience for people's daily routines of quickly retrieving packages. Summary of the Invention
[0003] To address the problems existing in the prior art, this invention provides an intelligent parcel pickup method and system for express stations based on image recognition. It has the advantages of achieving rapid location and visual guidance of express parcels by integrating technologies such as image positioning, text recognition, and semantic understanding of large language models, thereby improving the efficiency of parcel pickup at express stations and the user experience. It solves the problem that in the prior art, parcel pickup at express stations mostly relies on manual sorting or simple coding classification, without combining AI technology to achieve automated and intelligent parcel location and search, and thus cannot meet the high-efficiency pickup needs under high parcel volume.
[0004] This invention is implemented as follows: a smart parcel pickup method for post stations based on image recognition, comprising the following steps: S1, Form Recognition Model Training The LabelMe annotation tool was used to annotate the waybill area of the express delivery station. The annotated data was then input into the YOLOv8s model for training to obtain the "waybill recognition and localization image model", which is used to detect and locate the waybill area in the image. S2, Text Extraction Service Encapsulation Based on the above waybill positioning model, the image to be identified is segmented into several waybill regions; EasyOCR is called to perform text recognition on each waybill region, extracting express codes and various text information; the "waybill positioning + OCR text extraction" function is encapsulated into an independent "express search" service through MCP service, realizing the horizontal expansion of the system. S3, Semantic Understanding and Keyword Extraction The QWen3-8B large language model service is deployed through Vllm, and the semantic understanding capability of the large model is used to parse user input. The "express code" keyword in the user input is extracted through the Function-Call function. S4, Business Process Scheduling Using LangChain to call the large language model service and the "express delivery search" MCP service: First, obtain the express code from the user input, then call the MCP service to locate the waybill and extract the text from the parcel image at the station. After matching the target express delivery, generate an image with the target express delivery location marked. S5, Service Provision The above functions are encapsulated into an external service interface using FastAPI Python. After a user submits a pickup request via a mobile device, the system returns an image marking the location of the target package, thus providing intelligent pickup guidance.
[0005] A smart parcel pickup system for post stations based on image recognition, applicable to the aforementioned smart parcel pickup method based on image recognition, includes a waybill recognition module, a text extraction module, an MCP service module, a semantic understanding module, a business scheduling module, and an external service module.
[0006] As a preferred embodiment of the present invention, the waybill recognition module deploys a YOLOv8s waybill recognition and positioning image model to detect and locate the waybill area in the package image.
[0007] As a preferred embodiment of the present invention, the text extraction module integrates the EasyOCR tool to extract text from the located label area.
[0008] As a preferred embodiment of the present invention, the MCP service module is used to encapsulate the label recognition and text extraction functions as independent services, providing horizontal scalability.
[0009] As a preferred embodiment of the present invention, the semantic understanding module deploys the Vllm+QWen3-8B large language model service to realize semantic parsing of user input and extraction of express code.
[0010] As a preferred embodiment of the present invention, the service scheduling module calls the semantic understanding module and the MCP service module through LangChain to complete the process scheduling of the pickup request.
[0011] As a preferred embodiment of the present invention, the external service module encapsulates the service interface based on FastAPI to support the pickup request interaction on the mobile device.
[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: Effectively improve package pickup efficiency: AI automatically locates the package, replacing manual searching, shortening the user's pickup time and alleviating congestion at the pickup station during peak hours; Improve and optimize user experience: Visually mark the location of the target package to avoid misplacing and improve the convenience of package pickup; Effectively enhance system scalability: Core functions are encapsulated through MCP services to support horizontal system expansion in scenarios with multiple stations and high parcel volumes; Effective integration and synergy of technologies: For the first time in the field of express delivery pickup, technologies such as image positioning, OCR, large language model, and Function-Call are integrated to achieve full automation of the intelligent pickup process. Attached Figure Description
[0013] Figure 1 This is a schematic diagram of the item retrieval process provided in an embodiment of the present invention. Detailed Implementation
[0014] To further understand the invention's content, features, and effects, the following embodiments are provided, and detailed descriptions are given in conjunction with the accompanying drawings.
[0015] The structure of the present invention will now be described in detail with reference to the accompanying drawings.
[0016] refer to Figure 1 As shown in the figure, an embodiment of the present invention provides a smart parcel pickup method for post stations based on image recognition, which includes the following steps: S1, Form Recognition Model Training The LabelMe annotation tool was used to annotate the waybill area of the express delivery station. The annotated data was then input into the YOLOv8s model for training to obtain the "waybill recognition and localization image model", which is used to detect and locate the waybill area in the image. S2, Text Extraction Service Encapsulation Based on the above waybill positioning model, the image to be identified is segmented into several waybill regions; EasyOCR is called to perform text recognition on each waybill region, extracting express codes and various text information; the "waybill positioning + OCR text extraction" function is encapsulated into an independent "express search" service through MCP service, realizing the horizontal expansion of the system. S3, Semantic Understanding and Keyword Extraction The QWen3-8B large language model service is deployed through Vllm, and the semantic understanding capability of the large model is used to parse user input. The "express code" keyword in the user input is extracted through the Function-Call function. S4, Business Process Scheduling Using LangChain to call the large language model service and the "express delivery search" MCP service: First, obtain the express code from the user input, then call the MCP service to locate the waybill and extract the text from the parcel image at the station. After matching the target express delivery, generate an image with the target express delivery location marked. S5, Service Provision The above functions are encapsulated into an external service interface using FastAPI Python. After a user submits a pickup request via a mobile device, the system returns an image marking the location of the target package, thus providing intelligent pickup guidance.
[0017] A smart parcel pickup system for post stations based on image recognition, applicable to the aforementioned smart parcel pickup method based on image recognition, includes a waybill recognition module, a text extraction module, an MCP service module, a semantic understanding module, a business scheduling module, and an external service module.
[0018] Specifically, the waybill recognition module deploys a YOLOv8s waybill recognition and positioning image model to detect and locate the waybill area in the package image.
[0019] Specifically, the text extraction module integrates the EasyOCR tool to extract text from the located label area.
[0020] Specifically, the MCP service module is used to encapsulate the waybill recognition and text extraction functions into independent services, providing horizontal scalability.
[0021] Specifically, the semantic understanding module deploys the Vllm+QWen3-8B large language model service to realize semantic parsing of user input and extraction of express code.
[0022] Specifically, the business scheduling module calls the semantic understanding module and the MCP service module through LangChain to complete the process scheduling of pickup requests.
[0023] Specifically, the external service module encapsulates service interfaces based on FastAPI to support pickup request interactions on mobile devices.
[0024] Working principle of the invention: Using the visual positioning model YOLOv8s, specific blocks are identified and located. Then, EasyOCR is called to extract text from the image blocks and encapsulate it into a "Courier Search" MCP service. Then, the QWen3-8B model is deployed through vllm to provide semantic understanding and function-call services. While the large model service is called through LangChain to understand the user's input, the "number" of the courier that the user needs to find is obtained. Then, the "Courier Search" MCP service is called to obtain a complete image of the courier that is circled in red to mark the location of the target item.
[0025] This invention effectively improves package pickup efficiency by automatically locating packages using AI, replacing manual searching, shortening pickup time for users, and alleviating congestion at pickup stations during peak hours. It also enhances and optimizes user experience by visually marking the location of target packages to prevent accidental pickup and improve convenience. Furthermore, it strengthens system scalability by encapsulating core functions through MCP services, supporting horizontal expansion across multiple pickup stations and high-volume package scenarios. Finally, it effectively integrates and coordinates technologies, for the first time in the express pickup field, combining image positioning, OCR, large language models, and Function-Call technologies to achieve fully automated intelligent pickup processes.
[0026] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0027] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A smart parcel pickup method for post stations based on image recognition, characterized in that, Includes the following steps: S1, Form Recognition Model Training The LabelMe annotation tool was used to annotate the waybill area of the express delivery station. The annotated data was then input into the YOLOv8s model for training to obtain the "waybill recognition and localization image model", which is used to detect and locate the waybill area in the image. S2, Text Extraction Service Encapsulation Based on the above waybill positioning model, the image to be identified is segmented into several waybill regions; EasyOCR is called to perform text recognition on each waybill region, extracting express codes and various text information; the "waybill positioning + OCR text extraction" function is encapsulated into an independent "express search" service through the MCP service, realizing the horizontal expansion of the system. S3, Semantic Understanding and Keyword Extraction The QWen3-8B large language model service is deployed through Vllm, and the semantic understanding capability of the large model is used to parse user input. The "express code" keyword in the user input is extracted through the Function-Call function. S4, Business Process Scheduling Using LangChain to call the large language model service and the "express delivery search" MCP service: First, obtain the express code in the user input, then call the MCP service to locate the waybill and extract the text from the parcel image at the station. After matching the target express delivery, generate an image with the target express delivery location marked. S5, Service Provision The above functions are encapsulated into an external service interface using FastAPI Python. After a user submits a pickup request via a mobile device, the system returns an image marking the location of the target package, thus providing intelligent pickup guidance.
2. A smart parcel pickup system for post stations based on image recognition, characterized in that, The image recognition-based smart parcel pickup method for post stations as described in claim 1 includes a waybill recognition module, a text extraction module, an MCP service module, a semantic understanding module, a business scheduling module, and an external service module.
3. The intelligent parcel pickup system for post stations based on image recognition as described in claim 2, characterized in that: The waybill recognition module deploys a YOLOv8s waybill recognition and positioning image model to detect and locate the waybill area in the package image.
4. The intelligent parcel pickup system for post stations based on image recognition as described in claim 2, characterized in that: The text extraction module integrates the EasyOCR tool to extract text from the located area of the waybill.
5. The intelligent parcel pickup system for post stations based on image recognition as described in claim 2, characterized in that: The MCP service module is used to encapsulate the label recognition and text extraction functions into independent services, providing horizontal scalability.
6. The intelligent parcel pickup system for post stations based on image recognition as described in claim 2, characterized in that: The semantic understanding module deploys the Vllm+QWen3-8B large language model service to realize semantic parsing of user input and extraction of express code.
7. The intelligent parcel pickup system for post stations based on image recognition as described in claim 2, characterized in that: The business scheduling module calls the semantic understanding module and the MCP service module through LangChain to complete the process scheduling of pickup requests.
8. The intelligent parcel pickup system for post stations based on image recognition as described in claim 2, characterized in that: The external service module encapsulates service interfaces based on FastAPI, supporting pickup request interactions on mobile devices.