Underwater Waste Detection Using YoloV8 And Water Quality Assessment

The project addresses the issue of growing underwater waste in oceans and seas. It offers three solutions: YoloV8 Algorithm-based underwater waste detection, a rule-based classifier for aquatic life habitat assessment, and a Machine Learning model for water classification as fit for use or not fit. The first model was trained on a dataset of 5000 images, while the second model used chemical properties guidelines from US EPA and WHO. The third model was trained on a dataset with over 6 million rows, providing reliable water classification results.

Authors

• @ShivamVadalia
• @SiddharthChavan
• @KrunalBhere
• @FaizanNabi
• @ZuberPatel

Demo

Underwater Waste Detection Using YoloV8

Click here

Features

• Can detect underwater waste based on input images.
• Classifies water as potable or not based on chemical properties of water
• Classifies water as habitual for aquatic life or not.

Architecture of YoloV8

Tech Stack

• Python
• Dark Channel Prior Algorithm
• YoloV8(from Ultralytics)
• Xgboost-Classifier

Screenshots

Input Images:

Denoised Images After Running Dark Channel Prior Algorithm:

Output Images:

Datasets Used

• Underwater Trash Images
• Water Quality

Acknowledgements

• YoloV8 Training Notebook by Roboflow
• Architecture of YoloV8
• Single Image Haze Removal Using Dark Channel Prior