This repository contains code associated with the paper: Godbersen, P., Schanz, D. & Schröder, A. Peak-CNN: improved particle image localization using single-stage CNNs. Exp Fluids 65, 153 (2024)

We provide a python implementation of Peak-CNN using Tensorflow2/Keras in PeakCNN_funcs.py as well as a small testcase in testcase.ipynb showing training and application of the model to synthetic data.

If you find this work useful, please cite the coresponding paper referenced in Citation.bib

Peak-CNN itself depends only on Tensorflow and Numpy, the testcase notebook additionaly uses h5py and matplotlib. Their respective licenses can be found under:

• Tensorflow (LGPL): https://github.com/tensorflow/tensorflow/blob/master/LICENSE
• Numpy (custom): https://github.com/numpy/numpy/blob/main/LICENSE.txt
• h5py (BSD 3-Clause): https://github.com/h5py/h5py/blob/master/LICENSE
• Matplotlib (custom): https://github.com/matplotlib/matplotlib/blob/main/LICENSE/LICENSE