Coral Reef Image Classification Dataset for Deep Learning and Marine Biodiversity Monitoring

This dataset contains labelled underwater images of coral reefs prepared for supervised image classification and deep learning research. The images are organized into class-specific folders and divided into training, validation, and testing subsets using an 80:10:10 split. The dataset was developed to support the training, validation, and evaluation of computer vision models for automated coral reef image classification. It can be used to compare convolutional neural networks and transformer-based architectures, including ResNet, DenseNet, MobileNet, EfficientNet, ConvNeXt, and Vision Transformer models. . The dataset is suitable for research and educational applications in: • Coral reef image classification• Marine biodiversity monitoring• Underwater computer vision• Deep learning model benchmarking• Artificial intelligence for marine conservation• Environmental and ecological image analysis The underwater images may include variations in illumination, water clarity, viewing angle, image quality, background complexity, and reef appearance. These variations make the dataset useful for studying model robustness and generalization in realistic underwater environments. The files are provided as compressed archives while preserving the class-folder structure and the training, validation, and testing organisation. Dataset split:• Training set: 80%• Validation set: 10%• Testing set: 10% This dataset is intended for academic, research, and educational purposes. Model predictions should be validated by qualified marine biology or coral reef specialists before being used for ecological monitoring or conservation decision-making. Associated project:Coral Reef Image Classifierhttps://github.com/mohamedbadawi81/Coral-Reef-Image-Classifier