✈️ Aircraft Damage Detection — Deep Learning & Transformers

🔍 An end-to-end deep learning web application that detects aircraft damage from images using VGG16 transfer learning for classification and BLIP Transformer for AI-generated image captions and summaries. Built with Flask and served through a clean web interface.

---

📌 Overview

💡 Aircraft maintenance requires rapid and accurate damage detection. This project combines Computer Vision 🖼️ and Natural Language Processing 🧠 to:

• Classify aircraft images as Dent or Crack using a fine-tuned VGG16 CNN
• Generate natural language captions and summaries of the damage using the BLIP Transformer
• Visualize training performance curves (loss & accuracy) interactively

🔗 GitHub Repository: tanishcode-12/Aircraft-Damage-Detection

---

✨ Features

• 🛩️ Binary Damage Classification — Detects Dent or Crack from uploaded aircraft images
• 🤖 AI Image Captioning — BLIP Transformer generates natural language descriptions of the damage
• 📝 Damage Summarization — Produces a detailed summary of what the image shows
• 📈 Training Curves Visualization — Interactive loss & accuracy charts via Chart.js
• 🌐 Flask Web App — Upload images and get results instantly from your browser
• ⚡ Lazy Model Loading — Heavy ML models load only on first use for faster startup
• 🔒 File Validation — Accepts PNG, JPG, JPEG, WEBP (max 16 MB)
• 📊 Model Info API — Inspect architecture, input shape, optimizer, and hyperparameters

---

🧠 Model Architecture

🔹 Classification Model — VGG16

⚙️ Parameter	📝 Value
🏗️ Base Model	VGG16 (pretrained on ImageNet)
🖼️ Input Shape	224 × 224 × 3
🔒 Frozen Layers	All VGG16 conv layers
🧱 Custom Layers	Flatten → Dense(512, ReLU) → Dropout(0.3) → Dense(512, ReLU) → Dropout(0.3)
🎯 Output	Dense(1, Sigmoid)
📉 Loss	Binary Cross-Entropy
⚡ Optimizer	Adam (lr = 1e-4)
🏷️ Classes	Dent / Crack
🔢 Batch Size	32
🔁 Epochs	5

🔹 Captioning & Summary Model — BLIP

⚙️ Parameter	📝 Value
🤖 Model	Salesforce/blip-image-captioning-base
🧠 Framework	Hugging Face Transformers + PyTorch
📝 Task	Image Captioning & Damage Summarization

---

🗂️ Project Structure

✈️ Aircraft-Damage-Detection/
├── 🐍 app.py                          # Flask web app — routes & API
├── 🐍 aircraft_damage_detection.py    # Model training pipeline
├── 🌐 index.html                      # Frontend web interface
├── 📄 requirements.txt                # Python dependencies
├── 📁 uploads/                        # Uploaded images (auto-created)
└── 📝 README.md

---

🌐 API Endpoints

🔗 Endpoint	📡 Method	📝 Description
/	GET	🌐 Serves the main web interface
/api/classify	POST	🛩️ Classify image as Dent or Crack with confidence score
/api/caption	POST	📝 Generate BLIP caption and summary for uploaded image
/api/training-curves	GET	📈 Returns simulated training loss & accuracy history
/api/model-info	GET	🔍 Returns model architecture and hyperparameter details

---

⚙️ Installation

🧰 Prerequisites

• 🐍 Python 3.8+
• 📦 pip
• 🖥️ GPU recommended (for faster BLIP inference)

🪜 Steps

1. 📥 Clone the repository

git clone https://github.com/tanishcode-12/Aircraft-Damage-Detection.git
cd Aircraft-Damage-Detection

2. 📦 Install dependencies

pip install -r requirements.txt

3. ▶️ Run the application

python app.py

4. 🌐 Open in your browser

http://localhost:5000

---

🚀 Usage

1. 🟢 Launch the app using the steps above
2. 📤 Upload an aircraft image (PNG, JPG, JPEG, or WEBP — max 16 MB)
3. 🛩️ Click Classify to detect whether the damage is a Dent or Crack
4. 📊 View the confidence score and raw probability
5. 📝 Click Caption to get an AI-generated description and summary of the damage
6. 📈 View training curves to see model loss & accuracy over epochs

---

📦 Dependencies

📚 Library	🔧 Purpose
flask	🌐 Web framework & API routing
werkzeug	🔒 Secure file uploads
tensorflow / keras	🧠 VGG16 model & image preprocessing
torch	🔥 PyTorch backend for BLIP
transformers	🤗 Hugging Face BLIP model
Pillow	🖼️ Image loading & processing
numpy	🔢 Array operations
matplotlib	📊 Training curve generation

---

🤝 Contributing

🙌 Contributions are welcome! Here's how you can help:

1. 🍴 Fork the repository
2. 🌿 Create a new branch (git checkout -b feature/your-feature)
3. 💾 Make your changes and commit (git commit -m 'Add your feature')
4. 📤 Push to the branch (git push origin feature/your-feature)
5. 🔁 Open a Pull Request

✅ Please make sure your code is clean and well-commented.

---

👤 Author

Tanish — @tanishcode-12

---

⭐ If you found this project helpful, consider giving it a star on GitHub!