Quest's ThirdEye AI Vision Inspection couples computer vision with custom-trained DL models to detect defects faster, more accurately, and at a lower cost. This solution helps operations, plant and quality control managers in manufacturing optimize the way defect detection and quality control are managed in their lines. The solution works for very small items like semiconductors and, when more cameras are added, very large items like airplane wings. Operators live stream the camera views and see the defect detection with bounding boxes and comments on the screen, allowing them to validate the algorithm’s decision. The system performs detection at the edge allowing for real-time performance and easy scale-up. The edge logic is complemented in the cloud with code that performs continuous model improvements, notifies line managers of issues, and a full dashboard for defect analysis.

Key Features

• Custom AI computer vision algorithms running at the edge on an Azure Stack Edge device with NVIDIA GPUs that can do real-time defect detection from multiple camera streams
• Support multiple cameras and viewpoint to handle very large items (100 meter long airplane wings, train cars, turbine blades) in addition to small items
• Operator dashboard with real-time streaming and ability to validate defect detection
  

• Integration with Microsoft Teams for notifications and defect management

• Lower costs: Early, automatic defect detection leads to lower costs in recalled products as well as optimized inspection resources
• Higher quality: Defect detection not only improves production but also enables managers to train and improve operator performance
• Faster production: Manufacturing lines can move faster if human inspection is optimized, especially for large items like wind turbines