Inspiration

SafeSight AI — Agentic Industrial Safety Inspector

Inspiration

Industrial workplaces need timely safety inspections, but manually reviewing video footage can be slow and inconsistent. SafeSight AI explores how computer vision and cloud monitoring can help safety teams review potential incidents more efficiently.

What it does

SafeSight AI is a computer-vision-based industrial safety inspection application. Users upload workplace video for analysis, review detected events, inspect incident evidence, and access incident records through a web interface.

How we built it

We built the application with OpenCV 5, Python, FastAPI, React, and Vite. YOLO-based object detection supports visual analysis. AWS S3 provides evidence storage, while Amazon CloudWatch supports application logging and metrics.

Challenges we faced

We worked on connecting the video-analysis pipeline to the frontend, handling long-running requests, debugging API connectivity and CORS issues, and validating cloud integrations.

What's next

We plan to improve detection reliability, evaluate performance on varied workplace footage, expand incident classification, and strengthen the human review workflow.

Responsible use

SafeSight AI is a decision-support prototype, not a replacement for trained safety inspectors. Detection results require validation, and the system should be evaluated for false positives, false negatives, and performance across different environments.

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for SafeSight AI

Built With

• 5
• amazon
• amazon-web-services
• analytics
• artificial
• cloudwatch
• computer
• detection
• fastapi
• intelligence
• javascript
• learning
• machine
• object
• opencv
• python
• react
• s3
• video
• vision
• vite
• yolo