This video shows a live demonstration of our Senior Data Scientist, Scott Koval, testing his newly built model on its ability to detect if he is wearing a health mask. The models are written in Python using Jupyter notebook and run on a NVIDIA Jetson Nano Dev Kit.
The models are built to detect two scenarios:
Training data was generated within the notebook by taking example pictures of the subject demonstrating the two different classifications. A Computer Vision model was then trained on these data sets using a neural network. The video is showing the end result: trained models scoring Scott to determine whether he is wearing a health mask or not.
During serious disease outbreaks, businesses have to take precautions to keep their employees and customers safe. Wearing health masks can help prevent the spread of disease and save lives, which is why some retailers – such as Costco – make health masks mandatory during disease outbreaks. Using computer vision models like this to enforce health and safety standards can enable businesses to keep employees and customers safe.
In addition to image recognition like this, you can train the same type of model to recognize distance between people to ensure social distancing, another practice that has been shown to prevent the spread of disease. Staff can be alerted when people are too close together. Heatmaps of the business can be generated to show areas of higher risk so that staff can take action, such as redirecting customer flow or reorganizing that area, to lessen this risk. Forecasting can be used to predict times of day that have a higher risk of high traffic that impedes social distancing.
Computer vision data can be analyzed in real-time to send alerts to staff for immediate action when someone isn’t wearing a mask or two people get too close, and then the same data can be further analyzed and displayed on reports that give metrics on heatmaps and traffic forecasting to inform other, less immediate business actions and strategies.
Please fill out the form below.