Neural net ensemble for indexing of stainless steel 316L by directional reflectance microscopy

Description

Companion code and data to our paper "Encoding data into metals alloys using laser powder bed fusion".

Steps to reproduce

To test the code, we recommend installing Python > 3.6 and the dependencies listed in requirements.txt in a fresh environment. The main results from our study can be reproduced by executing the file qr_code_demo.py. The data folder contains the DRM training dataset we used. The saved_models folder contains one hundred trained MLP classifier models which can be reloaded and used for inference. These models were trained using different random initializations, which leads to minor differences in their output predictions.

Institutions

Categories

Licence