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.
TOP