D_six: Annotated Dataset for Floating Debris Detection

D_six is a curated dataset for the detection of floating debris in aquatic environments, designed to support the development of real-time object detection models such as YOLOv5.

The dataset is organized in standard YOLO format and includes:
- Training and validation images in `images/train` and `images/validation`
- Corresponding YOLO labels in `labels/train` and `labels/validation`
- A separate set of test images in the `test` folder

Classes include: plastic bottles, plastic drink container, cans, plastic take out , styrofoam, and plastic bags. Each label follows YOLO format: `[class x_center y_center width height]` with normalized coordinates.

Dataset Description:
The D_six dataset was developed for the detection and classification of floating debris in aquatic environments. It includes six representative debris categories collected from real-world aquatic scenes under varying lighting, occlusion, and wave conditions.

This dataset was used in the study:
“A2ANet: Real-Time Detection of Floating Marine Debris Using Atrous Convolution and Channel Attention” (submitted to Ecological Informatics).

Recommended citation:
Badams, B., (2025). D_six: Annotated Dataset for Floating Debris Detection. 

The dataset and its annotations are openly available under a Creative Commons Attribution License (CC-BY 4.0).