🎯 Underwater Object Detection in Sonar Imagery

📊 Project Highlights

• 🚀 Final Model: YOLOv8n (You Only Look Once)
• 🎯 Task: Object Detection, Bounding boxes and Lables & Classification
• 📈 Performance: State-of-the-art detection capabilities
• 🔄 Multiple Approaches: From Traditional ML to Deep Learning
• 🛠️ Comprehensive Pipeline: Data preprocessing to end to end Project

🌟 Key Features

• Multi-Model Comparison: Traditional ML vs Deep Learning approaches
• Automated Preprocessing: Standardized image processing pipeline
• Format Conversion: Handles multiple annotation formats (XML, YOLO txt)
• Performance Metrics: Detailed evaluation and visualization
• Modular Architecture: Easy to extend and modify

📂 Project Structure

1. Traditional ML Approach (traditionalML/)

flowchart LR
    A[Images + XML] -->|Feature Extraction| B[CSV Format]
    B -->|ML Training| C[Traditional Models]

Loading

2. Data Preprocessing (data_preprocessing/)

flowchart LR
    A[Raw Images] -->|Convert & Resize| B[Standardized Images]
    C[XML Annotations] -->|Update| D[Adjusted Annotations]
    B --> E[Processed Dataset]
    D --> E

Loading

• 🔄 Converts .bmp to .jpg
• 📐 Resizes images to 640x640
• 🎯 Updates bounding box annotations

3. Deep Learning Models

3.1 Faster R-CNN (objectDetectionDL_FASTER-RCNN/)

• Two-stage detector architecture
• Region proposal network
• Feature pyramid network integration

3.2 Image Classification (imageClassificationDL/)

• ResNet18 architecture
• Transfer learning approach
• Single-label classification

3.3 YOLOv8n - Final Model 🏆 (objectDetectionDL_YOLOv8n/)

flowchart LR
    A[Input Image] -->|YOLOv8n| B[Detection]
    B -->|Post-processing| C[Final Results]

Loading

📈 Model Performance

YOLOv8n Results

Performance Visualization

🔍 Detection Examples

🛠️ Setup & Usage

1. Environment Setup

   git clone <repository-url>
   cd <project-directory>

2. Data Preprocessing

   cd data_preprocessing
   python preprocess_images.py

3. Running YOLOv8n

   • Open objectDetectionDL_YOLOv8n/uatd_YOLOv8n.ipynb
   • Follow the notebook instructions for training and inference

📊 Results Analysis

The project demonstrates superior performance with YOLOv8n:

• High precision in object detection
• Real-time inference capabilities
• Robust performance across different scenarios

🔗 Dependencies

• Python 3.8+
• PyTorch
• Ultralytics YOLOv8
• OpenCV
• NumPy
• Pandas

📝 Notes

• Dataset is preprocessed using the pipeline in data_preprocessing/
• Annotations follow Pascal VOC XML format
• YOLOv8n provides the best balance of speed and accuracy

🌟 Future Improvements

• Model ensemble techniques
• Real-time video processing
• API deployment
• Mobile optimization

---

Object Detection Web Application

This is a web application that uses YOLOv8 for object detection in images. The application consists of a React frontend and a Node.js backend that interfaces with a Python script for model inference.

Prerequisites

• Node.js (v14 or higher)
• Python (v3.8 or higher)
• Your trained YOLOv8 model (uatd_yolov8_m-pytorch-default-v1.tar.gz)

Setup

1. Install Node.js dependencies:

npm install

2. Install Python dependencies:

pip install -r requirements.txt

3. Place your trained model file (uatd_yolov8_m-pytorch-default-v1.tar.gz) in the root directory of the project.

Running the Application

1. Start the backend server:

node server.js

2. In a new terminal, start the React development server:

npm start

3. Open your browser and navigate to http://localhost:3000

Usage

1. Click the "Upload Image" button to select an image from your computer
2. Once an image is selected, you'll see a preview
3. Click the "Detect Objects" button to run the object detection
4. The results will be displayed below the image, showing detected objects and their confidence scores

Project Structure

• src/App.js - React frontend application
• server.js - Node.js backend server
• detect.py - Python script for model inference
• requirements.txt - Python dependencies
• package.json - Node.js dependencies

Notes

• The application expects your model file to be in the root directory
• The backend server runs on port 3001 by default
• The React development server runs on port 3000
• Uploaded images are temporarily stored in the uploads directory and automatically deleted after processing

Made with ❤️ Tensors Ahmad, Hema & China