一种喷码字符检测的方法，装置及系统
By combining multiple pre-trained models with recognition and verification rules, inkjet printing detection samples are automatically extracted and generated, solving the problem of high accuracy in inkjet printing recognition in complex scenarios and achieving low-cost and high-efficiency inkjet printing detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ANHEUSER-BUSCH INBEV (WUHAN) BREWERY CO LTD
- Filing Date
- 2022-11-21
- Publication Date
- 2026-07-17
AI Technical Summary
Existing inkjet printing recognition technologies struggle to achieve high-precision detection in complex and diverse scenarios, especially in complex lighting and curved surface inkjet printing environments. Furthermore, data acquisition and annotation are difficult, traditional methods have high requirements for experimental conditions, deep learning methods require a large amount of labeled data, and GAN methods are unstable, failing to meet the low-cost, high-precision requirements of factories.
By combining multiple pre-trained models with recognition and verification rules, the system automatically extracts detection areas and generates labeled samples. Through simulation of various lighting and materials, it automatically generates training samples, reducing the precision requirement, improving the recall rate, and adapting to different scenarios.
It enables automatic sample labeling with little or no human intervention, improving the detection accuracy and adaptability of the inkjet printing detection system, reducing hardware and labor costs, and adapting to various complex scenarios.
Smart Images
Figure CN115909309B_ABST