VisionForge

Build Neural Networks Visually — Export Production Code

Design deep learning architectures with drag-and-drop. Export clean PyTorch or TensorFlow code instantly.

✨ What is VisionForge?

VisionForge is a visual neural network builder that lets you design complex deep learning architectures without writing code. Perfect for researchers, students, and ML engineers who want to rapidly prototype models.

• 🎨 Drag-and-drop interface — Build CNNs, LSTMs, ResNets visually
• ⚡ Automatic shape inference — No manual tensor dimension tracking
• 🔄 Multi-framework export — PyTorch or TensorFlow with one click
• 🤖 AI-powered assistant — Ask questions or modify your model with natural language
• ✅ Real-time validation — Catch architecture errors before export

🚀 Quick Start

Prerequisites

• Python 3.8+
• Node.js 16+
• Google Gemini API key (for AI assistant)

Installation

1. Clone and setup backend

git clone https://github.com/devgunnu/visionforge.git
cd visionforge/project

# Install Python dependencies
pip install -r requirements.txt

# Setup environment
cp .env.example .env
# Add your GEMINI_API_KEY to .env

# Initialize database
python manage.py migrate

# Start Django server
python manage.py runserver

Backend runs at http://localhost:8000

2. Setup frontend

cd project/frontend

# Install dependencies
npm install

# Start dev server
npm run dev

Frontend runs at http://localhost:5173

3. Open your browser Navigate to http://localhost:5173 and start building!

🎯 How It Works

┌─────────────┐    ┌──────────────┐    ┌──────────────┐    ┌─────────────┐
│  Drag & Drop │ →  │  Configure   │ →  │   Validate   │ →  │   Export    │
│    Blocks    │    │  Parameters  │    │ Architecture │    │    Code     │
└─────────────┘    └──────────────┘    └──────────────┘    └─────────────┘

1. Add layers from the sidebar (Conv2D, LSTM, Dropout, etc.)
2. Connect blocks to define your model's data flow
3. Set parameters by clicking on any block
4. Validate your architecture with built-in checks
5. Export production-ready code for PyTorch or TensorFlow

📦 Available Layers

Core Layers Input / Output Linear (Fully Connected) Conv1D / Conv2D / Conv3D Flatten, Reshape Activation & Regularization ReLU, Sigmoid, Tanh, Softmax Dropout, BatchNorm Layer Normalization

Pooling MaxPool2D, AvgPool2D AdaptiveAvgPool2D Recurrent & Sequence LSTM, GRU Embedding Operations Add, Concatenate Custom layers

💡 Example Architectures

Simple CNN Classifier

Input → Conv2D → ReLU → MaxPool2D → Flatten → Linear → Softmax → Loss

ResNet-style Skip Connection

Input → Conv2D → BatchNorm → ReLU ┐
                                   ├→ Add → ReLU → Output
       Input ────────────────────→┘

LSTM Sequence Model

Input → Embedding → LSTM → Dropout → Linear → Output → Loss

🛠️ Tech Stack

Frontend	React 19 • TypeScript • Vite • Tailwind CSS
Canvas	ReactFlow • Zustand • Framer Motion
Backend	Django • Python • SQLite
AI	Google Gemini API
UI	Radix UI • shadcn/ui

⌨️ Keyboard Shortcuts

Shortcut	Action
Ctrl+Z	Undo
Ctrl+Y	Redo
Delete	Remove selected block

📖 Documentation

• Quick Start Guide
• AI Chatbot Setup
• Export Format Specification
• Node Architecture

🐛 Troubleshooting

Backend not starting?

cd project
python manage.py migrate
python manage.py runserver

Frontend build errors?

cd project/frontend
rm -rf node_modules package-lock.json
npm install
npm run dev

CORS issues? Make sure both servers are running (Django on 8000, Vite on 5173)

🤝 Contributing

Contributions are welcome! Feel free to open issues or submit PRs.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

📄 License

Please check the LICENSE file.

---

Ready to build AI models faster?

Get Started • View Docs • Report Issues

Made by devgunnu