• 

  scan this and see all the details about our project

🌊 Detection of Microplastics Entering the Human Body Through Water 🌟 Inspiration Behind the Project

Plastic pollution has emerged as one of the most critical environmental challenges of the modern world. Scientific reports indicating the presence of microplastics in drinking water, marine ecosystems, food, and even human bloodstreams motivated me to explore this issue further. The realization that humans may unknowingly ingest thousands of microplastic particles annually inspired the development of a technological solution to detect these contaminants early.

The core idea of this project originated from a simple but impactful question:

Can we design an intelligent system that detects microplastics in water before they enter the human body?

This curiosity led to the integration of Artificial Intelligence, image processing, and full-stack web technologies to create an automated detection and monitoring system.

🎯 Objectives of the Project

The primary goals of the project are:

To detect microplastic particles present in water samples.

To classify particles using Machine Learning techniques.

To estimate contamination levels and potential health risks.

To provide a digital platform for monitoring and analysis.

The concentration of microplastics in water is calculated using:

𝐶

𝑁 𝑉 C= V N ​

Where:

𝐶 C = Concentration of microplastics (particles per liter)

𝑁 N = Number of detected particles

𝑉 V = Volume of water sample (liters)

🛠️ How I Built the Project

The project was developed using a combination of hardware analysis, AI modeling, and MERN stack web development.

1️⃣ Data Collection and Sample Preparation

Water samples were collected from various sources such as tap water, bottled water, and local water bodies. The samples were filtered using microfilter paper to capture microscopic particles. The residue obtained was then dried for imaging.

2️⃣ Image Acquisition and Processing

Microscopic images of filtered particles were captured using a digital microscope. These images were processed using Python and OpenCV techniques such as:

Noise reduction

Thresholding

Edge detection

Segmentation

Mathematically, preprocessing can be represented as:

3️⃣ Machine Learning Model Development

A Convolutional Neural Network (CNN) was trained to classify particles into:

Microplastic

Non-plastic

The convolution operation is represented as: Feature=(Input∗Kernel)+Bias

The model learned features such as texture, shape, and color variations that distinguish plastics from organic or mineral particles.

4️⃣ Full-Stack Application Development (MERN)

To make the system user-friendly and accessible, a web platform was developed using the MERN stack:

MongoDB → Data storage

Express.js & Node.js → Backend APIs

React.js → Frontend interface

The platform allows users to:

Upload sample images

View detection results

Monitor contamination levels

Analyze historical data

5️⃣ Risk Level Estimation

The contamination risk is categorized using threshold logic:

This helps users understand potential exposure risks easily.

📚 What I Learned

This project provided knowledge in multiple domains:

🔹 Technical Skills

Image processing using OpenCV

Deep learning model development (CNN)

Full-stack development using MERN

REST API integration

Database design and management

🔹 Research Skills

Understanding environmental datasets

Problem analysis and solution design

Scientific documentation and experimentation

🔹 Personal Development

Team collaboration

Problem-solving under constraints

Debugging and optimization techniques

⚠️ Challenges Faced

Several challenges were encountered during development:

1️⃣ Dataset Availability

Obtaining real microplastic datasets was difficult due to limited public resources and laboratory access.

2️⃣ Particle Similarity

Non-plastic particles such as sand, fibers, and organic matter often resembled microplastics, reducing classification accuracy.

3️⃣ Small Object Detection

Microplastics are extremely small, requiring high-resolution imaging and careful preprocessing.

4️⃣ Integration Complexity

Connecting the AI model with the MERN stack backend required API integration and data format optimization.

5️⃣ Model Overfitting

Due to limited training data, the model initially overfitted. This was addressed using:

Data augmentation

Regularization

Cross-validation

🌍 Impact and Future Scope

This project contributes toward environmental safety and public health awareness by providing an affordable detection method.

Future improvements include:

IoT-based real-time monitoring

Mobile application integration

Larger training datasets

Government water quality monitoring integration

Built With

• css
• html
• json
• react