AI-based image analysis and computer vision

Business Unit »Test and Reliability Solutions«

We design, build, and run production-grade computer-vision pipelines for metrology and inspection in semiconductor manufacturing. Our models close the loop to process engineering by delivering robust, explainable metrics that lower cost and enable process monitoring and optimization.

We combine proven computer-vision (CV) techniques with state-of-the-art deep learning (DL) to fit the problem at hand. Where established algorithms excel, we use them for speed and stability, where they fall short, we extend them with tailored models − from classical CV to advanced DL. For very specific challenges, we create custom architectures and post-processing steps that embed process knowledge and allow fine-tuning for corner cases.

This hybrid strategy yields explainable outputs and robust performance across process corners. The result is the automation of formerly manual or semi-automatic inspection and metrology with measurable cost savings. Customers gain higher accuracy, reliability, and consistency − plus clear, actionable metrics that feed directly into optimization and root-cause analysis.

Use case examples

Automated via dishing and long-scan analysis in AFM images

Deterministic CV pipeline for explainable, reliable results, including GUIs and automated image acquisition and storage.

Carbon nanotube detection and segmentation in AFM images

Customized deep-learning model deriving morphological features such as orientation, length, and density distributions.

Defect detection CD-SEM images

Deep-learning–based segmentation to detect areas with post-etch residue (PER) and a via misalignment offset (ALO) after lithography and etching steps.