🦅 FALCON-Net

Welcome to FALCON-Net! 🚀

An interactive Streamlit playground for visualizing and testing the robustness of Siamese and Prototypical neural networks under adversarial attacks.

🌐 Live Demo: FALCONNet

---

🎯 What is FALCON-Net?

FALCON-Net is your go-to app for exploring how cool neural networks (Siamese & Prototypical) handle sneaky adversarial attacks! Draw, attack, visualize, and learn—all in one place. Perfect for students, researchers, and the just-plain-curious. 😎

---

✨ Features

• 🎨 Draw & Attack: Doodle your own character, unleash attacks (FGSM, PGD), and see the chaos unfold with heatmaps and metrics!
• 🗂️ Pick & Attack: Choose from a gallery of multilingual characters, attack them, and compare before/after results.
• 📊 Metrics Dashboard: Dive into interactive charts showing accuracy, robustness, and more.
• 🧬 Siamese Network Explorer: Peek inside each layer, visualize feature maps, and compare embeddings.
• 🧑‍🤝‍🧑 Prototypical Network Explorer: Play with few-shot learning, support/query sets, and see how prototypes work.

---

🛠️ Tech Stack

• Frontend/UI: Streamlit 🎈
• Deep Learning: TensorFlow/Keras 🧠
• Visualization: Matplotlib, Pillow 📷
• Data: Sample character images (assets/) 🖼️

---

🚦 Get Started

1. Clone the Repo

git clone https://github.com/your-username/FALCON-Net.git
cd FALCON-Net

2. Install the Magic

We recommend a virtual environment! 🪄

pip install -r requirements.txt

3. Launch the App

streamlit run app.py

Open your browser to http://localhost:8501 and let the fun begin! 🎉

---

🗂️ App Map

• app.py — Main hub, navigation, and page routing
• draw_page.py — Freehand drawing & attack fun
• select_page.py — Pick a character & attack
• metrics_page.py — Performance dashboards
• siamese_page.py — Siamese network explorer
• prototypical_page.py — Prototypical network explorer
• attacks.py — Adversarial attack code (FGSM, PGD)
• image_utils.py — Image helpers
• assets/ — Character images

---

🕹️ How to Play

• Use the sidebar to jump between pages 🧭
• Follow the on-screen prompts to draw, attack, and explore
• Try different attacks and see how the models react
• Have fun and learn something new! 🤓

---

🚀 Deployment

The app is live on Streamlit Cloud: FALCONNet

Want your own version? Fork this repo and connect to Streamlit Community Cloud in minutes!

---

📜 License

Licensed under the GNU General Public License v3.0. See LICENSE for details.

---

🙏 Acknowledgements

• Inspired by the amazing world of adversarial robustness & few-shot learning
• Built with Streamlit, TensorFlow/Keras, and open-source magic ✨

---