In this project, I used the technique of transfer learning, which is to adopt algorithm from available models and froze all the layers before the ouput layer. The origin model I chose is Xception, and the modified model provides an accuracy of 0.9837 in the final test.
Zion-Ma/AOI-on-Wafer-Defect
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.