Abnormality detection method, device and equipment for automobile die casting image and storage medium
An anomaly detection method based on multi-scale feature extraction and fusion of automotive die-casting images solves the problems of fatigue and subjectivity in manual visual inspection, achieving high-precision automated anomaly detection and improving product quality and inspection efficiency.
CN120852869APending Publication Date: 2025-10-28GUANGDONG HONGTEO ACCURATE TECH ZHAOQING CO LTD
View PDF 0 Cites 0 Cited by
Patent Information
- Application Number
- CN202510966979.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-14
- Publication Date
- 2025-10-28
AI Technical Summary
Technical Problem
In the current technology, the detection of anomalies in automotive die-cast parts relies on manual visual inspection, which suffers from problems such as fatigue, strong subjectivity, and low detection accuracy, making it difficult to guarantee the consistency and reliability of product quality.
Method used
A feature extraction module and a feature fusion module are used to extract and fuse multi-scale features from images of automotive die-cast parts. Combined with an anomaly detection module, a model is trained to generate an anomaly probability map, thereby achieving automated anomaly detection.
Benefits of technology
It improves the accuracy and consistency of anomaly detection, reduces labor costs, and enhances product quality reliability and detection efficiency.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN120852869A_ABST
Abstract
The invention relates to the field of image anomaly detection, in particular to an anomaly detection method, device and equipment for an automobile die casting image and a storage medium, and aims to extract features of a normal image of an automobile die casting and a constructed abnormal image of the automobile die casting. Feature fusion is carried out on the obtained normal images of the automobile die castings and the feature extraction images corresponding to the corresponding abnormal images of the automobile die castings to carry out anomaly detection, anomaly probability graphs corresponding to the normal images of the automobile die castings are obtained, model training is carried out in combination with abnormal mask images constructed based on the normal images of the automobile die castings, and the abnormal images of the automobile die castings are obtained. The trained model can effectively extract the abnormal information in the image, and the abnormal detection precision is improved.
Need to check novelty before this filing date? Find Prior Art