Underwater Pipeline Detection Dataset (YOLO Format)

Dataset Overview

The dataset offers YOLO‑format data for underwater pipeline detection, comprising training and validation sets, their respective annotation files, and a class definition file (classes.txt).

Dataset Contents

• Training set (train): 6,396 images with annotations.
• Validation set (val): 1,575 images with annotations.
• Class file (classes.txt): Defines all target classes for the detection task.

Directory Structure

project/
├── images/
│   ├── train/      # training images
│   ├── val/        # validation images
├── labels/
│   ├── train/      # training annotations
│   ├── val/        # validation annotations
├── classes.txt     # class definition file
└── README.md       # project description

Data Format Description

• Image files: Supported formats include JPG, PNG, etc.
• Annotation files:
  • Each image has a same‑named .txt file.
  • Each line represents an object in the format: <class_id> <x_center> <y_center> <width> <height>.
    • class_id: class index starting at 0.
    • x_center and y_center: normalized object center coordinates relative to image size.
    • width and height: normalized bounding‑box dimensions relative to image size.

Example Files

Sample Image and Annotation

Image

images/train/sample.jpg

Annotation

labels/train/sample.txt content example:

0 0.5 0.5 0.4 0.3
1 0.7 0.8 0.2 0.1