ISense is an end-to-end full-process AI industrial quality inspection platform developed by Hanswell based on deep learning to solve the core problems of industrial complex defect detection and engineering management, realize the rapid migration of cross-product model, and meet the visual application of multiple sub-industry scenarios.
Five core advantages to promote the upgrading of industrial quality inspection
1, continuous learning
Based on the characteristics of deep learning, according to new data, learn new product forms and reduce catastrophic forgetting.
2. Rapid deployment and replacement of production line
Support one-click migration learning, and realize the rapid deployment of multi-model and multi-type workpieces.
Verify the prototype of visual scheme, and shorten the time period of visual project integration and deployment.
3. Learning with noise
Improve the learning robustness under high noise ratio
Automatic correction and elimination of abnormal data
4. Automatic defect sample generation
Defect sample generation to meet the learning needs of small samples.
Enhance the richness of samples and reduce the long tail effect of data.
5. Lower sample dependence
The number of samples required is less than 200, and the detection accuracy is more than 99%.
Unsupervised learning, only good products are needed to meet the needs of anomaly detection.
Super-resolution image, non-uniform sampling