Overview

This repository contains the python scripts to build and train a two-headed LeNet architecture to regress the orientation vector of suspended objects from two perpendicular views. This methodology is intended to measure the orientation of single suspended particles rotating in viscous shear flows. Given the impossibility to obtain labeled training data for this application, the model here presented was intended to be trained on synthetic data.

Organization

Custom models and data loaders can be found in src Run the main.py file on the training data to

Data

Training data as well as experimental recordings of the rotation of axis-symmetric particle suspended in a confined shear flow in the viscous and small-inertia regime can be found at the following repository

License

This repository is licensed under a cc-by-4.0 license.
You are free to use, share, and adapt this dataset for any purpose, provided you give appropriate credit to the original author.

Citation

If you use this repository in your work or are inspired by our approach, please cite our articles:

[1] Di Giusto, D. and Bergougnoux L. and Guazzelli, É., 2025, Orientation of flat bodies of revolution in shear flows at low Reynolds number, Journal of Fluid Mechanics, 1017, p.A41.

[2] Di Giusto, D. and Bergougnoux, L. and Marchioli, C. and Guazzelli, É., 2024. Influence of small inertia on Jeffery orbits, Journal of Fluid Mechanics, 979, p.A42.