🎓 Transfer Learning Course with PyTorch

A beginner-friendly deep learning transfer learning course using PyTorch framework, perfect for getting started with computer vision.

📚 Course Overview

This course teaches you transfer learning techniques from scratch through hands-on image classification projects using pre-trained models:

• Lesson 1: Environment Setup & Verification
• Lesson 2: Data Exploration & Analysis
• Lesson 3: ResNet18 Transfer Learning
• Lesson 4: ResNet50 Transfer Learning
• Lesson 5: EfficientNet-B0 Transfer Learning
• Lesson 6: EfficientNet-B3 Transfer Learning
• Lesson 7: MobileNet-V2 Transfer Learning

🌸 Dataset Information

Flowers-102 Dataset

This course uses the Oxford Flowers-102 dataset for all transfer learning experiments:

• Total Images: 8,189 flower images
• Classes: 102 different flower categories
• Split:
  • Training: ~1,020 images
  • Validation: ~1,020 images
  • Test: ~6,149 images
• Image Size: Variable (resized to 224×224 for training)
• Source: Oxford Visual Geometry Group

Sample Categories: Alpine Sea Holly, Anthurium, Bee Balm, Bird of Paradise, Bolero Deep Blue, Bougainvillea, Camellia, Canna Lily, Canterbury Bells, Carnation, and 92 more beautiful flower species.

Why Flowers-102?

• Perfect balance of complexity and manageability
• High-quality, diverse images
• Ideal for demonstrating transfer learning benefits
• Automatic download through torchvision
• Well-established benchmark dataset

🚀 Quick Start

For Windows Users

1. Double-click start_notebook.bat
2. The script will automatically:
   • Set up the Python environment (first time only)
   • Install necessary packages
   • Launch Jupyter Notebook in your browser

For macOS / Linux Users

1. Open terminal
2. Run the startup script:

   chmod +x start_notebook.sh  # First time only
   ./start_notebook.sh

Manual Setup (Advanced)

If you prefer manual control, you can use python setup.py to create the environment, and then activate it manually:

# Create env
python setup.py

# Activate (Windows)
transfer_learning_env\Scripts\activate

# Activate (Mac/Linux)
source transfer_learning_env/bin/activate

# Start Jupyter
jupyter notebook

Learning Path

1. Start with Lesson 1: Open lessons/lesson1_environment_setup/setup_check.ipynb to verify your environment.
2. Follow the progression: Lessons 1-7 build upon each other.
3. New to Notebooks?: See QUICKSTART.md.

📖 Course Structure

transfer-learning-course/
├── start_notebook.bat           # One-click launch for Windows
├── start_notebook.sh            # One-click launch for Mac/Linux
├── setup.py                     # Environment setup script
├── requirements.txt             # Package dependencies
├── lessons/
│   ├── lesson1_environment_setup/ # Environment verification
│   ├── lesson2_data_exploration/  # Data exploration
│   ├── lesson3_resnet18/          # ResNet18 hands-on
│   ├── lesson4_resnet50/          # ResNet50 hands-on
│   ├── lesson5_efficientnet_b0/   # EfficientNet-B0 hands-on
│   ├── lesson6_efficientnet_b3/   # EfficientNet-B3 hands-on
│   └── lesson7_mobilenet_v2/      # MobileNet-V2 hands-on

💻 System Requirements

• Python: 3.8 or higher
• OS: Windows 10+, macOS 10.14+, Ubuntu 18.04+
• Memory: At least 4GB RAM (8GB recommended)
• Storage: At least 2GB free space
• GPU: Optional but recommended (CUDA-compatible)

🎯 Learning Objectives

By completing this course, you will learn:

1. Understand transfer learning principles and applications
2. Master PyTorch deep learning framework
3. Use pre-trained models for image classification
4. Compare different model architectures and performance
5. Optimize training processes and hyperparameters
6. Deploy and apply trained models

📊 Model Performance Comparison

Expected performance on Flowers-102 dataset:

Model	Parameters	Training Time	Accuracy
ResNet18	11.7M	~20 min	~85%
ResNet50	25.6M	~35 min	~88%
EfficientNet-B0	5.3M	~25 min	~90%
EfficientNet-B3	12.2M	~40 min	~92%
MobileNet-V2	3.5M	~15 min	~83%

Performance may vary based on hardware and training settings

🔧 Troubleshooting

Installation Issues

1. Python Version: Ensure Python 3.8+
2. Network Issues: Check internet connection
3. Permission Issues: Run in administrator mode
4. Reinstall: Delete transfer_learning_env folder and run setup again

Cannot Start Jupyter?

1. Ensure virtual environment is activated
2. Run jupyter --version to check installation
3. Try jupyter notebook --no-browser to start

Slow Training?

1. Use GPU: Install CUDA version of PyTorch
2. Reduce batch size: If memory insufficient
3. Use smaller models: Like MobileNet-V2

🔬 Advanced Features

• Automatic dataset download: No manual data preparation needed
• Progress tracking: Real-time training monitoring
• Model comparison: Side-by-side architecture analysis
• Visualization tools: Training curves and predictions
• Export capabilities: Save and load trained models

📚 Documentation

📑 Documentation Index:

• DOCUMENTATION_INDEX.md - Complete documentation index

⚙️ Setup Guides:

• INSTALLATION_GUIDE.md - Installation Guide
• QUICKSTART.md - Quick Start Guide

📞 Getting Help

• Issues: Submit on GitHub
• Discussions: Check course discussion board
• Documentation: Read detailed lesson instructions and tutorials above

🎉 Start Learning

Everything is ready! Open lessons/lesson1_environment_setup/setup_check.ipynb to begin your deep learning journey!

---

Happy Learning! 🚀

💡 Tip: Follow lessons in order, each includes detailed explanations and code examples.