Handwash Compliance Monitor

Inspiration

There is currently no real-time system in hospitals or restaurants to monitor hand hygiene compliance, leading to poor enforcement and accountability. This lack of oversight contributes to the spread of infections, higher healthcare costs, employee absences, and increased environmental waste due to excessive use of PPE and sanitation supplies.

What it does

It uses real-time video and YOLO-based object detection to track handwashing quality and duration, ensuring all WHO-recommended steps are followed. With pose estimation and BOT-SORT tracking, it accurately monitors hand and joint movements throughout the process.

How we built it

We built our project by first sourcing a labeled dataset from Roboflow, which contained 1,064 images categorized into six sequential handwashing steps. This structured data allowed us to train a YOLOv8 model - we chose for its real-time capabilities, accuracy, and ease of use - for step-by-step hand hygiene detection. Using Roboflow’s tools, we prepped and trained the model with a custom configuration. We then integrated the trained model with OpenCV to capture live webcam input and detect handwashing steps in real time. To track progress, we implemented logic to count how often each step appeared and confirmed success based on both step completion and time spent washing. We enhanced the accuracy of our detection with BOT-SORT, a tracking algorithm that follows single objects across frames. Finally, we built a simple Flask web interface to stream the video feed and provide instant visual feedback when handwashing is completed correctly.