Treffer: CNN Parameter Tuner 3.0.0

Title:
CNN Parameter Tuner 3.0.0
Authors:
Kottilapurath Surendran, Gokul, orcid:0000-0002-1360-, Mokros, Martin
Publisher Information:
Zenodo
Publication Year:
2025
Collection:
Zenodo
Subject Terms:
Convolutional Neural Network, CNN, Deep Learning, Image Classification, Machine Learning, Neural Network Tuning, CNN Parameter Optimization, Data Preprocessing, Image Augmentation, Jupyter Notebook Generator, Python Script Generator, Model Training, Train-Test Split, Computer Vision, AI Tool, Custom CNN, Educational Tool, Automated ML Workflow, Reproducible Research, Tree Species Identification, Forestry Applications, hyperparameter tuning, AI Research and Development, Deep Learning Frameworks, AI Model Training Tools, Model Fine-tuning, AI Development Tools
Document Type:
E-Ressource software
Language:
English
ISSN:
1574-9541
Relation:
https://zenodo.org/communities/euraf/; https://zenodo.org/communities/reforest/; https://zenodo.org/communities/eu/; https://zenodo.org/records/15348641; oai:zenodo.org:15348641; https://doi.org/10.5281/zenodo.15348641
DOI:
10.5281/zenodo.15348641
Availability:
https://doi.org/10.5281/zenodo.15348641
https://zenodo.org/records/15348641
Rights:
Creative Commons Attribution 4.0 International ; cc-by-4.0 ; https://creativecommons.org/licenses/by/4.0/legalcode
Accession Number:
edsbas.A6A2D552
Database:
BASE

Weitere Informationen

Greetings and welcome to CNN Parameter Tuner 3.0.0! This innovative application serves as your gateway to the exciting world of Convolutional Neural Network (CNN) parameter tuning. It empowers you to explore and optimize CNN algorithms with precision and ease, unlocking the full potential of image classification tasks. The tool offers a wide array of preprocessing options, including normalization, standardization, and various augmentation techniques. You can also experiment with different train-test-validation ratios, custom CNN architectures, epochs, batch sizes, activation functions, optimizer choices, and loss functions. Additionally, the application provides the convenient option to generate a Python script (Jupyter Notebook) based on the selected parameter values for future use or further customization. Please ensure that your images are organized into separate folders, with each folder name representing a class label. Upon loading the dataset, the application will automatically detect and assign these folder names as the class labels for the classification task. Kindly note that the software may take approximately one minute to appear. We appreciate your patience! New Feature: We are pleased to introduce the ability to generate Python scripts based on your chosen parameters. Please be aware that the results obtained in the generated notebook may differ slightly from previous runs, due to the inherent randomness in training processes such as data splitting and weight initialization.