In the field of machine vision defect detection, AI, or artificial intelligence, mainly refers to an automatic detection algorithm with deep learning as the core.
Based on deep neural network, the model is trained and verified through supervised learning, which takes the marked defective and good pictures as the collection. The trained data were then used to detect the unknown images.
As a training set, the more defect images and good images, the more comprehensive distribution, the wider coverage of defect types, the final detection effect will be better.
AI visual detection and screening equipment advantages
- Solve the defect detection problem that is difficult to program using traditional visual algorithm;
- It can adapt to relatively more interference factors and reduce the requirement for consistency of products;
- Can adapt to relatively poor lighting conditions, thus compatible with more product types;
- When introducing new products, there is no need to rewrite the core algorithm, and the import speed is faster;
- No or only need to set a few parameters, can meet the detection requirements, reduce the difficulty of debugging;
- Autonomous new software allows users to complete algorithm model training and deployment of new artifacts by themselves.
Traditional algorithm:
The traditional algorithm mainly uses the width and continuity characteristics of thread bright edge to detect.
AI deep learning algorithm:
AI deep learning algorithm detects through image annotation, execution training, verification process, model testing and other steps.