Monsight

• 

  Monsight_Hero

• 

  Live_Monsight_Demo

• 

  Live_Navigation_Demo

• 

  The Agentic Loop

• 

  Final Monsight

InspirationMonSight is an AI-powered assistive navigation and safety agent designed to help blind and visually impaired users understand their surroundings and move more safely through changing environments.

MonSight combines real-time computer vision, voice interaction, navigation context, and safety reasoning in an agentic loop: Perceive → Understand → Reason → Decide → Act → Verify.

The working prototype uses a camera to detect objects and estimate their position, proximity, and confidence. A Safety & Risk tool then prioritizes hazards and produces actions such as STOP, SLOW DOWN, CAUTION, VERIFY, or CONTINUE. Users can interact through voice commands, including navigation requests such as going to a destination. The prototype also includes simulated route information and a medicine-label OCR tool that reads printed medicine information such as name, strength, and expiry.

The system is built locally using Python, Ultralytics YOLO, OpenCV, Vosk, Tesseract OCR, pyttsx3, NumPy, and Pillow, avoiding paid cloud APIs.

MonSight is designed as a decision-support system, not a replacement for human judgment or certified mobility technology. Its future direction is richer GPS/maps, indoor navigation, personalized assistance, and wearable smart-glasses integration.

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Monsight

Built With

• ai
• computer-vision
• machine-learning
• numpy
• opencv
• pillow
• python
• pyttsx3
• tesseract-ocr
• ultralytics-yolo
• vosk