Semi-conductor Image Classification

Jan 30, 2020
  • Helped Prof. YAO Yuan and the Nexperia company with the Nexperia semi‐conductor classification problem;
  • Distinguished broken chips from normal chips by using deep learning and image classification methods;
  • Compared the difference of perform between different models, e.g., ResNet20, ResNet56;
  • Tried transferred learning methods. It turned out that a ResNet56 model transferred from a pretrained ResNet20 model could perform better, than simply training a ResNet56 model directly.
Kelley Kan HUANG
Kelley Kan HUANG
Engineer (PI)

My research interests include distributed robotics, mobile computing and programmable matter.