🧠 AiDoor

AiDoor is a joke SaaS webapp that uses a self-trained neural network to detect whether an image contains a door. It answers with maximum authority:

yes.
Or, if it’s unsure:
hmm. idk.

Built entirely for learning, laughs, and practice in React, Python, and AI workflows using ChatGPT + Cursor.

---

🚀 Features

• 🖼️ Drag and drop image upload UI
• 🧠 Python neural network (PyTorch)
• 🧪 Confidence-based prediction
• 🤖 Feedback-based model improvement (WIP)
• 🛠️ Clean dev workflow using ChatGPT XML + Cursor
• 🗂️ Modular project structure (frontend/backend split)

---

🧱 Project Structure

AiDoor/
├── frontend/ # ReactJS UI (upload, results)
├── backend/ # Flask backend
│ ├── app.py # Main API (predict endpoint)
│ ├── analyze.py # Image processing + inference
│ ├── model/
│ │ ├── train_model.py # Training script
│ │ └── door_model.pth # Trained model (in .gitignore)
│ └── data/ # Images for training (door/ + not_door/)
├── readmes/ # Deep documentation (ai.md, workflow.md, etc.)
├── .gitignore
└── README.md # You're here

---

💻 How to Run Locally

1. Backend (Python)

cd backend
python -m venv venv && source venv/bin/activate
pip install -r requirements.txt
python app.py
Frontend (React)

bash
Copy
Edit
cd frontend
npm install
npm run dev
Access App

Open http://localhost:5000 or your React dev server.

🧠 How It Works
User uploads or drags an image

Image is sent to the backend via Flask API

analyze.py:

Resizes image

Converts to tensor

Feeds it into a PyTorch CNN model

Outputs confidence score

If confidence > 0.5 → yes.

Else → hmm. idk.

📦 AI Model Training
Run train_model.py to retrain the model from scratch

Uses a basic CNN on:

bash
Copy
Edit
backend/data/
  ├── door/
  └── not_door/
Saved as model/door_model.pth (ignored in Git)

✍️ Workflow Philosophy
Inspired by the B2025 project

🧩 Plan in XML-style prompts with ChatGPT

🧠 Build/code using Cursor

📚 Document deeply in /readmes

📦 Keep clean separation of logic (frontend/backend/model)

📎 Future Plans
Add user feedback to retrain model

Add loading animations

Build “AiDiode” spin-off (joke SaaS #2)

Make a “Model Confidence Dashboard”

Add TTS ("yes." in robotic voice)

⚠️ Disclaimer
This is a joke side project intended to practice web app development, AI model training, and workflow discipline. Do not actually use this to detect doors in real applications. Please.

📸 Screenshot (Coming Soon)
🧑‍💻 Built With
ReactJS (Vite)

Flask

PyTorch

HuggingFace (optional)

ChatGPT (XML prompting)

Cursor (IDE)

yaml
Copy
Edit