Vision Access - Image Description Add-on
This add-on enables NVDA users to describe graphical content on screen using AI. You can use either Ollama
(local) or Google Gemini AI.
Features
- Describe the visual of navigator object or focused object
- Support for Ollama (local installation) or Gemini (cloud)
- Percentage-based progress notifications
- Ask follow-up questions about the image
- Turkish and English interface support
Keyboard Shortcuts
| Shortcut |
Function |
| NVDA+Shift+E |
Describe the navigator object's visual |
| NVDA+Shift+G |
Describe the focused object's (graphic, image) visual |
Settings
To access settings: NVDA Menu → Preferences → Settings → Vision Access
Provider Selection
- Ollama (Local): Local AI model running on your computer
- Gemini (Google AI): Google's cloud-based AI service
Ollama Settings
- Ollama URL: Ollama server address (default: http://localhost:11434)
- Model: Vision model to use (e.g., llava, moondream)
Gemini Settings
- API Key: API key obtained from Google AI Studio
- Model: Gemini model selection (gemini-2.5-flash, gemini-2.5-pro, etc.)
Requirements
For Ollama
- Ollama must be installed on your computer (ollama.ai)
- A vision-capable model must be downloaded (e.g.,
ollama pull llava)
For Gemini
- Free API key from Google AI Studio
- Internet connection
Usage
- Navigate to the visual you want to describe (image, graphic, screen area)
- Press NVDA+Shift+E or NVDA+Shift+G
- Listen to progress notifications (Percent 0, 25, 50, 75, 100)
- When you hear "Description ready", the result dialog opens
- Use "Ask More" field to ask follow-up questions
Troubleshooting
Ollama connection error: Make sure Ollama is running (ollama serve)
Gemini API error: Ensure your API key is valid and quota is not exceeded
Developer
Sarper Arıkan
Email: sarperarikan@gmail.com
Web: sarperarikan.net