Skip to content
Shruthi.
← Back to projects

Guard AI.

Computer vision that spots household hazards and explains them.

  • React
  • Flask
  • Python
  • YOLOv8
  • GPT API
View code ↗

The problem

Households with young children or elders needing care carry everyday hazards that are easy to walk past. Object detection alone only names what it sees — a list of detected objects is not advice, and a caregiver still has to work out what to actually do about it.

Approach

  1. Used YOLOv8 object recognition to detect potentially hazardous objects in a home environment in real time.

  2. Passed detections to the GPT API to turn raw labels into specific safety recommendations and precautions for each object found.

  3. Built the frontend in React for an interactive interface that surfaces risks as they are identified.

  4. Powered the backend with Flask, handling API requests and coordinating the YOLOv8 and GPT stages.

Outcome

  • Detection results arrive as actionable precautions, not just object labels.
  • Real-time analysis suits the walkthrough case — scanning a room and getting feedback immediately.
  • Aimed squarely at families caring for children or elders, where the cost of a missed hazard is highest.

In full

A home safety application that leverages YOLOv8 object recognition to detect potentially hazardous objects in a home environment. The application integrates with the GPT API to provide safety recommendations and precautions for the detected objects. This tool is designed to assist families with children or elders requiring care by identifying risks and suggesting actionable safety measures in real-time. The frontend is built using React for an interactive user interface, while the backend is powered by Flask to handle API requests and integrate YOLOv8 and GPT functionalities.