/MicroECT-3D

Primary LanguagePythonBSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

🧪 µECT

Microscale 3-D Capacitance Tomography with a CMOS Sensor Array.

Overview

Screenshot

Input and Output formats

  • Input is a 3-D matrix of capacitance measurements of size (m x n x r) = (20, 10, 5)
  • Output is a 3-D volume of size (200, 100, 5) -> (200um, 100um, 50um)

Download 3-D Dataset

Download sample 3-D dataset:

./data/datasets/download.sh

The data will be downloaded to data/datasets/3-D/dataset

Training

  python3.7 train_3d.py --config <experiment-config> --exp_name <experiment-name>

For example, run the following to train on the downloaded dataset:

  python3.7 train_3d.py --config config/experiments/3d/07112023.yaml --exp_name 07112023_3D > experiments/07112023_3D.log 2>&1 &

This by default outputs the logs to experiments/

Inference

  python3.7 evaluate_3d.py --config  <experiment-config> --model <path-to-model> --output_dir <output-dir>

For example, run the trained model:

  python3.7 evaluate_3d.py --config  config/experiments/3d/07112023.yaml --model experiments/07112023_3D/best_model.pth 

License

BSD 3-Clause License. See LICENSE.