R&D
Provide high quality, high speed inspection machines
R&D
Provide high quality, high speed inspection machines
DEEP LEARNING
"What if a machine learns by experience?"
R&D team of Pixel targets on integrating Artificial Intelligence with machine vision technologies to automate inspection processes that currently require extensive manual labor.
Motivation
Traditionally Inspection systems use rule-based algorithms to detect defects. However, due to the high cost of missing true defects and the productivity issue, inspection machines exhibit rather a high rate of false detection. Most of times, over 90% of defects that are detected by the inspection machines are false. These false defects are then verified and filtered by human operators. Our R&D team is developing ways of integrating DEEP LEARNING methods to the existing inspection systems in order to automatically remove false alarms.
Benefits of using DEEP LEARNING
- Continuously trained for new defects without the need of change in Algorithm.
- Sensitivity can be controlled by adjusting the level of training.
- False alarms can be reduced without the help of human by proper training.
To implement, need to consider the following steps
- Classify images for training.
- Feed the classified images to our system and start training.
- The system automatically trains itself based on the given images.
- Export the neural network to our inspection system.