FLSea Stereo Dataset

The Data provides a comprehensive underwater stereo imaging collection from the shallow Mediterranean waters off the coast of Israel. Designed for enhanced depth mapping and object size estimation, this dataset contains four distinct sets of stereo image data captured at 10 frames per second. It includes calibration data for precise intrinsic and extrinsic camera parameter adjustments. Each set is meticulously categorized based on location, ensuring easy access and application for underwater research and AI training. Additionally, it features depth maps generated using Agisoft Metashape for improved accuracy in spatial analysis.

Image Diversity

The dataset contains thousands of stereo image pairs taken from various depths and locations in the ocean. These images showcase diverse underwater scenes, including coral reefs, marine flora, and fauna, offering a comprehensive view of underwater ecosystems.

Detailed Annotations

Each image pair is accurately labeled with depth information, camera parameters, and environmental conditions, providing crucial data for precise 3D modeling and analysis. These annotations are vital for training and testing machine learning algorithms.

Applications

The Data is invaluable for developing underwater navigation systems, enhancing marine biology studies, and advancing autonomous underwater vehicle technology. By utilizing this dataset, researchers can significantly improve the accuracy of underwater imaging and contribute to the exploration and conservation of marine environments.