Industrial safety / Computer vision
Computer Vision for Industrial Safety
Using visual understanding to support safety monitoring in industrial environments.
Problem statement
Recognize safety-relevant events in camera data so they can inform an operational response.
My contribution
Development of pose-detection and scene-interpretation methods for automated safety monitoring.
Technical challenges
Domain-level considerations, without implementation-specific details.
- Interpreting human activity from visual measurements.
- Distinguishing safety-relevant events from ordinary activity.
- Connecting a detection with a meaningful review or response.
Methodology
Combine pose detection and generative AI to identify person-down events and connect detection to automated alarms.
- Pose-based representations of human activity in images and video.
Conceptual architecture
Visual input
Pose estimation
Event interpretation
Alert and review
Technologies and methods
- Computer Vision
- Pose Detection
- Generative AI