What is Deep Learning?
Deep learning is a subset of machine learning and therefore artificial intelligence. It employs artificial neural networks, with numerous intermediate layers between the input layer and the output layer, thereby forming an extensive internal structure.
The deep learning algorithm is trained with large data sets, from which rules are generated that can then be applied to other data sets. In this way, complex sets of rules can be created that do not have to be programmed by hand
Defect detection with artificial intelligence
Using artificial intelligence, the proven inspection performance of the MBJ electroluminescence tester has once again significantly improved.
A pre-trained neural network reliably detects typical defects such as micro-cracks, dark areas and soldering defects, while the pseudo defect rate is once again significantly reduced.
MBJ also provides the profound hyper parameter optimization for:
- Best computing performance
- Minimum processing power
- And fastest feature extraction
No time-consuming optimization
Remarkably, the deep learning method is very robust to variations in the cell material.
- The time-consuming and error-prone optimization of the image processing parameters is completely eliminated.
- Only a few more sample images are required for optimization. With relatively little effort, additional defect classes can be added according to customer requirements.
MBJ offers image processing with neural networks for all electroluminescence inspection systems in its product range.
Benefits of Deep Learning in MBJ EL Tester
Improved inspection performance
Detects typical defects such as micro-cracks, dark areas and soldering defects, while the pseudo defect rate is once again significantly reduced.
Flexible
New defect types can be easily trained with just a small number of samples and without software change.
Easy
Our support team will be happy to help you train the networks to achieve the best results.
We can also provide you with all the necessary tools to train the networks by yourself on site without any special knowledge.