The Vision-Based Safety Team is committed to creating safer makerspaces by leveraging computer vision and object detection to identify potential hazards—particularly those related to personal protective equipment (PPE) and improper clothing. Our current focus is utilizing lightweight, efficient vision models that can run directly on a Raspberry Pi, and integrating these models with strategically placed cameras. We aim to detect issues such as missing safety goggles, improper footwear, and other non-compliant attire in real-time.
Key areas of research include optimizing vision-based software for limited hardware resources, refining training datasets for diverse makerspace environments, and designing non-intrusive safety alerts for students. This approach helps prevent accidents and fosters a culture of proactive safety awareness among students and staff.
Looking ahead, our long-term goal is to deploy a permanent hardware-software system in campus makerspaces that seamlessly identifies and addresses safety concerns without disrupting normal activities.
