Ships Image Dataset

Data Description

• Total Images: The dataset comprises 8,506 high-quality images.
• Annotation Format: Ship annotations are provided in YOLO v5 PyTorch format, making it convenient for users to implement YOLO-based object detection models.
• Pre-processing Applied:
  • Auto-orientation of pixel data to ensure uniformity (EXIF-orientation stripping applied).
  • Resized to a consistent dimension of 600×416 pixels (stretching applied to fit this size).
• Augmentation Techniques:
  • Each source image has three augmented versions.
  • 50% probability of a horizontal flip to introduce variability.
  • Random Gaussian blur applied with a blur radius ranging from 0 to 3.75 pixels, enhancing the dataset’s versatility.

Classes

The dataset includes images of ships classified into the following categories:

• Aircraft Carrier
• Bulkers
• Car Carrier
• Container Ship
• Cruise
• Destroyer (DDG)
• Recreational
• Sailboat
• Submarine
• Tugboat

Inspiration

The primary goal of this dataset is to explore and enhance AI’s capabilities in detecting and classifying different types of ships in images. By leveraging this dataset, researchers and developers can create sophisticated models that can accurately identify various ship types, contributing to advancements in maritime surveillance, transportation, and security.

Use Cases

• Maritime Surveillance: Enhancing monitoring systems for coastal and international waters.
• Navigation and Safety: Assisting in the development of autonomous navigation systems for ships.
• Environmental Monitoring: Tracking and managing maritime traffic to minimize environmental impact.
• Military Applications: Improving the identification and tracking of naval vessels for defense purposes.

Acknowledgements

Special thanks to the Roboflow team for their significant contributions in sharing this valuable data, making it accessible for researchers and developers to drive innovations in the maritime domain.