🐶 Dog Breed Classification Desktop App

This is a standalone desktop application built using Tkinter and TensorFlow, capable of identifying dog breeds from images. The app uses a trained .h5 deep learning model and supports creating .exe and .msi installers via cx_Freeze or PyInstaller.

🚀 Features

✅ Offline desktop app (no Python installation required)
🖼️ Upload any dog image (JPG, PNG, JPEG)
🔍 Predicts dog breed with confidence score
📂 Simple folder structure and user-friendly GUI
🛠️ Create .exe or .msi installers with cx_Freeze
⚠️ Error handling for missing models or files

🧠 Model Info

Trained on 31 dog breeds including:


Akita Inu, German Shepherd, Labrador Retriever, Poodle, Pug, Golden Retriever,
Siberian Husky, Bulldog, Saint Bernard, Rottweiler, and more.

Input Image Size: 224 x 224
Format: dog_breed_classifier.h5 (Keras model)

📁 Folder Structure


DogBreedApp/
│
├── app.py                  # Main GUI application file
├── dog\_breed\_classifier.h5 # Trained TensorFlow model
├── class\_indices.json      # Label-to-classname mapping
├── setup.py                # cx\_Freeze build script
├── dog.ico                 # (optional) Icon file

🛠 Requirements

Install these Python packages:

pip install -r requirements.txt

requirements.txt

tensorflow
numpy
pillow
cx-Freeze

🖥️ Running the App

python app.py

📦 Creating Executable (.exe or .msi)

✅ Using cx_Freeze

Create setup.py
Build:

python setup.py build

For MSI Installer:

python setup.py bdist_msi

This generates:

.exe in build/
.msi installer in dist/

🧊 Packaging Notes

Use resource_path() for model/JSON so PyInstaller or cx_Freeze can bundle files.
Include .h5 and .json in include_files in setup.py.
Works without requiring Python installation.

📸 Screenshots

📜 License

This project is open-source and available for educational and personal use.

👨‍💻 Author

Developed by Jayasimma D


---

### ✅ Bonus Tips:
- Save it as `README.md` in your root folder.
- Add screenshots to a `/screenshots` folder if you plan to upload this to GitHub or Share.
- Replace `[Your Name]` with your actual name or GitHub username.