CNNsTruffle — Truffle Crack Detection with Deep Learning

Code and image data for detecting truffle cracks in soil images with convolutional neural networks (CNNs). The models help truffle hunters tell cracks caused by growing truffles apart from other cracks in the field.

Related paper

Rasul, A. (2025). Automated Truffle Crack Detection Using Deep Learning and Machine Learning. Preprints. doi:10.20944/preprints202508.0941.v2

Repository contents

Path	What it is
code/truffie__27_09_2025_three_high_performed_models.ipynb	Google Colab notebook that trains and compares transfer-learning CNNs (VGG16, MobileNetV2, EfficientNetV2) and an SVM classifier
data/	The labelled soil-crack images used for training and testing

How to run

1. Clone the repository:

   git clone https://github.com/Azad77/CNNsTruffle.git
   cd CNNsTruffle

2. Open the notebook in code/ in Google Colab.
3. Copy the data/ folder to MyDrive/data in your Google Drive (the notebook mounts Drive and reads the images from there), then run the cells in order.

Libraries used: TensorFlow/Keras, scikit-learn, NumPy, pandas, Pillow, Matplotlib and joblib.

Citation

If you use this code or data, please cite the paper above.

Contact

Dr. Azad Rasul — azad.rasul@soran.edu.iq