Water Quality Assessment

Plastic waste in aquatic environments has a catastrophic effect on the surrounding ecosystem as well as the economic and environmental conditions of the area and beyond. Our research proposes a deep learning based framework of detecting instances of semantic waste including but not limited to plasti

Project Title

Water Quality Assessment

Project Area of Specialization

Artificial Intelligence

Project Summary

Plastic waste in aquatic environments has a catastrophic effect on the surrounding ecosystem as well as the economic and environmental conditions of the area and beyond. Our research proposes a deep learning based framework of detecting instances of semantic waste including but not limited to plastic bottles, wrappers, containers, tubes and bags. We plan to use a sizeable and publicly-available dataset of actual debris in open-water locations to train the network and obtain accurate and real time results.
 

Project Objectives

Motivation for the undertaking of this research project revolves around the growing threat of debris in water channels, such as improperly recycled packaging, discarded trash or miscellaneous waste. Particularly in Pakistan, the municipal authorities and the environmental regulatory bodies have failed to enforce their respective regulations to stop the dumping of garbage on such a mass scale into the rivers, leading to a massive 164,332 tonnes of plastic debris in the Indus River. 

This project aims to use object detection in order to identify the plastic waste present on the surface of these water channels in order to monitor the level of pollution spatially as well as temporally. This will lead to the source of this pollution being narrowed down and will pave the way for a feasible and efficient solution. This will affect not only rivers but also go a long way in removing plastic from oceans as well.

Project Implementation Method

We propose a camera-based solution that can be installed near riverbanks in order to monitor the amount of pollution by object detection of litter through a deep learning framework. The first step would be construction of the dataset, which would include the following classes

• Plastic waste such as cans, cups, bags, bottles and wrappers
• Biological organisms such as fish, plants and other biological detritus

Adoption of publicly available datasets such as GINI is also an option. The model will be designed to find all plastic debris, regardless of type. Next, an object detection network will be chosen in order to optimize performance without compromising on speed. This network will be trained according to our object detection metrics and will receive data from the camera module placed in the field. 

The camera module is an important feature of the project. It needs to be placed in a protected environment, with a heat proof and dust proof casing and provided with a reliable data connection such as 4G. Power supply can not be guaranteed due to the remoteness of the location hence a solar powered solution would be more reliable. 
 

Benefits of the Project

The benefits of the project are far-reaching. The most significant benefit is the increase in awareness for the proper disposing of plastic trash. Data obtained from the monitoring of plastic levels in water channels could be analysed in order determine and possibly tackle the source of the problem.

Moreover, this project can act as a stepping stone for future work in the area. Any dataset created from scratch or modified would be usable for future endeavours of the same vein. This work could be extended to the streets and roads of particular cities instead of water channels as well. Other projects such as the trash cleaning robot on the Chicago river could arise for Pakistani rivers as well as a result of the contributions and benefits of this project.

Technical Details of Final Deliverable

1. Reusable object detection dataset for plastic waste
2. Deep learning network trained with plastic waste classes
3. System that delivers real time data on the levels of plastic on water channels
    

Final Deliverable of the Project

HW/SW integrated system

Type of Industry

Agriculture , Health

Technologies

Artificial Intelligence(AI)

Sustainable Development Goals

Clean Water and Sanitation

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com