Contact Us
INF-ADC
Automatic Defect Classification System

Overview

As IC manufacturing processes grow more complex, manual defect classification can no longer keep pace with industry demands. By leveraging high-performance GPUs, AI-assisted defect classification and signature detection methods have been introduced to greatly reduce costs and speed up yield enhancement.

Based on AI technologies, like image classification and computer vision, Semitronix offers a comprehensive solution called INF-ADC to automatically classify defects that are generated in different steps and equipment during the wafer manufacturing processes. INF-ADC provides an intuitive user interface and does not require machine leaning expertise. Users can develop and apply ADC models by using INF-ADC without a single line of code.
Overview
Main Features

undefined

Advantages

Ease of use:

  • Smooth data deduplication, cleaning, and labeling

  • AI-assisted labeling

  • Code-free development for model generation

Wide range of applications:

  • Support any process step

  • Support any image format

  • Support customized requirement

High classification accuracy:

  • Built-in cutting-edge deep learning models pre-trained with massive semiconductor data

  • Precise and efficient, with an average recognition accuracy of 98.5%

  • Support online data feedback for continuous model optimizatio

High system performance:

  • Fast concurrent inference for up to 20ms/image

  • Deployment of multiple instances and machines to ensure high reliability

  • Real-time display of ADC model statistics to report generation

  • Monitoring of model status

 

XML 地图