Several methods came to mind.  The first thing needed was the integrating it with software. We decided to focus on Zoom since its the most used platform. After searching the internet, found that there is no SDK that works from user side. 

Next option was to run a python automation script on command. 

Next, the feedback, computer vision seemed like the best option but 1. it was expensive, 2. some people might not like having a camera (privacy concerns) therefore, we decided to go with a simple PIR sensor. 

used Arduino nano because that was available, and we did not need a lot of inputs and outputs.