Deep Diffraction

A series of deep convolutional neural networks to automatically analyze position averaged convergent beam electron diffraction patterns (PACBEDs) for automatically processing big, 4D STEM data.

1. Auto-align raw PACBEDs without the need for pretreating the data.
2. Measure the sample thickness and tilt.
3. Hybrid CNN + least square fitting (LSF) for more general PACBED analysis.

Neural network and simulation files are about 2GB size. Please contact Prof. James LeBeau ([email protected]) for files.

If you find this software package is useful, please cite: Weizong Xu, James M. LeBeau, A Deep Convolutional Neural Network to Analyze Position Averaged Convergent Beam Electron Diffraction Patterns, arXiv:1708.00855, 2017