Underwater Plastic Pollution Detection

Enhanced underwater images of ocean debris for plastic pollution detection

Source
Kaggle
Region
Global
Format
jpg
Access
Open access
Themes
Plastics, Air Quality & Pollution, Oceans & Sea Level, Machine Learning & AI

Get the data from Kaggle

About this dataset

These images depict underwater scenes marred by discarded garbage and debris polluting the oceans. Employing a preprocessing method known as Dark Prior Channel, the dataset has been enhanced to amplify image contrast, facilitating the identification and detection of the debris.

The file structure is organized into three primary directories:

Train Directory:

Contains 3,628 images along with their associated labels.

Valid Directory:

Comprises 1,001 images paired with their respective labels.

Test Directory:

Encompasses 501 images along with their labels.

The labels encompass values representing the bounding box coordinates (e.g., x-coordinate of the top-left corner, y-coordinate of the top-left corner, width, height) along with a class label.

Number of Class - 15

Class Names - ['Mask', 'can', 'cellphone', 'electronics', 'gbottle', 'glove', 'metal', 'misc', 'net', 'pbag', 'pbottle', 'plastic', 'rod', 'sunglasses', 'tire']

Browse sustainability datasets on SusanHub