Our team is dedicated to developing automated solutions that enhance workspace safety and regulations compliance. Our research focus lies at the intersection of machine learning, and real-time data processing. We are developing intelligent systems capable of detecting and ensuring the correct use of Personal Protective Equipment (PPE) with computer vision.
Our primary project is an “edge computing” safety solution to enforce goggle compliance in the metal room. We are engineering a standalone device, powered by a Raspberry Pi, that uses a custom-trained YOLOv8 object detection model. The system processes a live video feed locally to identify personnel at the entrance. If the model determines a user is not wearing safety goggles, it provides immediate, real-time feedback by activating a visible warning, such as a red LED strip or a message on a small display. The impact is a proactive, automated safety check that prevents eye injuries before an individual enters the work environment.
To achieve optimum accuracy, our team utilizes Convolutional Neural Networks (CNNs) and dedicated object detection architecture (YOLO) as the core of our detection system. A significant portion of our work involves meticulous data collection and processing—building and labeling strong, diverse datasets that enable our models to perform reliably under the required environmental conditions.
Looking ahead, we hope to broaden our system’s capabilities for implementation in diverse spaces and more challenging conditions, while also making it capable of detecting other PPE items. This can be done through further improvement of our training and machine vision practices. We also continue to optimize performance on edge hardware, such as small Single Board Computers(e.g. Raspberry Pi). Ultimately, we envision a future where intelligent vision acts as a vigilant partner in creating an infallibly safe environment for all users.
