Automated Image-Based Measurement Tool

What it does

Transistors are essential components in integrated circuits, where their performance depends on their structure and material composition. Recently, conventional silicon-based CMOS logic devices have encountered technical and physical limitations. Consequently, many research teams are exploring new semiconductor materials to advance the field.
Our project aimed to develop an artificial intelligence model to assess the quality of channel samples obtained during the exfoliation step, which is critical in the fabrication process of semiconductor devices based on novel materials. Traditionally, these channel samples are observed under an optical microscope or Scanning Electron Microscope (SEM), and their width is measured manually using scientific image measurement software.
However, we have designed a website that utilizes the Gemini API and image processing techniques to automate this process. Users can upload target photos to the API and immediately receive information about the width of the channels and the suitability of the samples for device fabrication. Plus, We also created and trained our own datasets for segmentation models based on the UNet architecture. We believe this innovation will greatly benefit research teams working on nano devices based on novel semiconductor materials, providing them with significant convenience.