We are excited to announce the release of CNN Parameter Tuner 3.2.0, a powerful yet accessible application developed within the ReForest project to make deep learning tools more user-friendly and adaptable for agroforestry research and beyond.
The CNN Parameter Tuner is designed to help researchers, students, and practitioners optimise Convolutional Neural Networks (CNNs) without needing extensive coding experience. Users can load their own image datasets and fine-tune a wide range of parameters, including network architecture, learning rate, batch size, epochs, activation functions, optimizers, and loss functions. The tool also enables easy export of workflows as Python Jupyter Notebooks, ensuring reproducibility and seamless integration into larger research pipelines.
CNN Parameter Tuner has already proven its robustness in research applications. A recent study published in Ecological Informatics demonstrated how CNNs tuned with this tool outperformed classical algorithms in challenging image classification tasks, such as identifying tree species from bark images, with accuracy rates exceeding 90%. Combined with its user-friendly interface and fully open, the software is quickly gaining traction across disciplines, with hundreds of downloads already recorded.
By combining advanced machine learning with accessibility, the CNN Parameter Tuner supports innovation across ecology, agroforestry, and education and represents a key step toward making artificial intelligence both practical and inclusive.
Dutch 
English 
German 
French 
Spanish 
Danish 
Hungarian 
Czech 
Bulgarian