🌊 Intelligent Underwater Object Detection Using Deep Learning

An AI-powered web application for underwater image enhancement and object detection using YOLOv11, TensorFlow, Grad-CAM, and Streamlit.

---

📖 Overview

Underwater images often suffer from poor visibility, low contrast, color distortion, and haze, making object detection challenging.

This project presents an intelligent underwater object detection system that enhances underwater images using a deep learning-based dehazing model before performing object detection with YOLOv11. To improve model transparency, Grad-CAM is integrated to visualize the regions influencing predictions.

The application provides an intuitive Streamlit interface with image enhancement, object detection, explainability, and an AI chatbot.

---

✨ Features

• 🌊 Underwater Image Dehazing
• 🎯 Underwater Object Detection using YOLOv11
• 🔥 Grad-CAM Explainability
• 🤖 Gemini AI Chatbot
• 🔐 User Authentication
• 📥 Download Enhanced Images
• ⚡ Automatic & Manual Dehazing Modes
• 🖥️ Interactive Streamlit Web Interface

---

🏗️ System Architecture

                 Underwater Image
                        │
                        ▼
             Image Quality Analysis
                        │
         ┌──────────────┴──────────────┐
         │                             │
         ▼                             ▼
  Image Dehazing                  Original Image
         │                             │
         └──────────────┬──────────────┘
                        ▼
                 YOLOv11 Detection
                        │
                        ▼
              Bounding Box Prediction
                        │
                        ▼
              Grad-CAM Visualization
                        │
                        ▼
             Streamlit Web Application
                        │
                        ▼
               Gemini AI Assistant

---

🛠️ Tech Stack

Technology	Purpose
Python	Programming Language
YOLOv11	Object Detection
TensorFlow	Image Dehazing
PyTorch	Grad-CAM
OpenCV	Image Processing
Streamlit	Web Application
SQLite	Authentication
Gemini API	AI Chatbot

---

📂 Project Structure

Underwater-Object-Detection-Using-Deep-Learning
│
├── app.py
├── README.md
├── requirements.txt
├── .gitignore
│
├── notebooks
│   ├── image_dehazing.ipynb
│   └── yolo11_gradcam.ipynb
│
├── model
│   └── README.md
│
├── trained_model
│   └── README.md
│
└── sample_images

---

📊 Dataset

The project is trained using an underwater object detection dataset containing marine organisms and underwater debris.

Example Classes

• Fish
• Crab
• Eel
• Starfish
• Marine Plants
• ROV
• Plastic Bottles
• Plastic Bags
• Nets
• Ropes
• Containers
• Cups
• Wreckage
• Other Marine Debris

---

⚙️ Installation

Clone Repository

git clone https://github.com/uttejkaramala/Underwater-Object-Detection-Using-Deep-Learning.git

cd Underwater-Object-Detection-Using-Deep-Learning

Install Dependencies

pip install -r requirements.txt

---

📁 Required Models

This repository does not include the trained models due to GitHub size limitations.

Place the following files before running the application.

YOLO Model

model/
└── best.pt

TensorFlow Dehazing Model

trained_model/
├── saved_model.pb
└── variables/
    ├── variables.data-00000-of-00001
    └── variables.index

---

▶️ Run Application

streamlit run app.py

---

🚀 Workflow

1. Upload an underwater image.
2. Select Automatic or Manual Dehazing.
3. Enhance the image using the dehazing model.
4. Detect underwater objects using YOLOv11.
5. Visualize model attention with Grad-CAM.
6. View detected objects and confidence scores.
7. Download enhanced images or interact with the AI chatbot.

---

🔮 Future Improvements

• Real-time underwater video detection
• Marine species classification
• Object tracking
• Edge device deployment
• Autonomous underwater vehicle integration
• Multi-language chatbot support

---

👨‍💻 Author

Uttej Karamala

B.Tech in Artificial Intelligence

Java Full Stack Developer | Machine Learning Enthusiast

---