-
-
Mask Detector Working
-
Mask Detector Working
-
Social Distancing Detector Working
-
Social Distancing Detector Working
Inspiration
COVID-19 has grown to be a massive pandemic that has affected everyone. Whether that be causing people to be socially inactive, experience health problems, or face hardships in families, relationships and jobs. The pandemic needs to end. That's where we come in!
What it does
Our group has created an application, which would be integrated with security cameras, and used in public areas to ensure COVID protocols. These protocols include being socially distanced from one another and wearing masks. Wondering how? Through computer vision, machine learning, and neural networks, our program detects the distance between two individuals (making sure they are far away from one another) and checks if they are wearing masks. If either parameter is false, a message on the screen is being created to alert the person using the application. While this is handy, our software also creates a video of the footage captured by the camera, to be reviewed later if needed.
How we built it
The language used was Python, and the libraries were OpenCV, Keras, Numpy, and TensorFlow. They all assisted in enhancing the computer vision and accuracy of the application.
Challenges we ran into
There were several challenges that we ran into. First of all, we struggled to understand how to detect the distance between two individuals. We understood how to do this though, by looking at the OpenCV documentation (using the Histogram of Oriented Gradients feature). After overcoming this hurdle, we faced another issue, of a greater level. It was integrating mask detection into the code. Eventually though, with trial and error, the problem was fixed, and we got a working code!
Accomplishments that we're proud of
We were proud to have worked together to create an efficient software that serves a good use for the public. Moreover, we were also especially impressed with the accuracy of the mask detection.
What we learned
We obtained a better grasp of the Open-CV library and learned how to effectively work in a collaborative environment. Furthermore, having never worked with TensorFlow before, we also developed a grasp of this computer vision library, as the project progressed.
What's next for COVID Safety Device
We hope to remove any remaining inconsistencies with the social distancing detection. Additionally, we also hope to develop the mask dataset so we can train the model to determine whether someone is wearing a mask or not based on a side profile (rather than just a front profile).
Log in or sign up for Devpost to join the conversation.