We are creating a project to detect the occupancy of empty chairs vs occupied chairs, we have trained our model to only detect a empty chair and a chair with a person in it, but, we have run into some issues:
Are we going about this the best way? Think of how AI garage systems operate, when someone takes a parking space it is occupied and updates it somehow? We have no clue how to go about that/tackle that, I have been researching for hours. Our best idea so far is to just have it ONLY detect an empty chair and a chair with a person in it, is this a good idea or is there a better one? And how would we send this to a website to show which chairs are occupied and which chairs are not. I’m assuming we would have to give each chair an ID#, but we do not know how to do that either. If anyone can steer me in the right direction I would really appreciate it.
Or, could we create some sort of edge detection box around the chair and when something enters it, it changes its status to occupied? Is that even possible? Thank you in advance.
Maybe you can try to train a detector for chairs as well as a classifier for empty/occupied status.
Since you have a detector for a chair, you can apply some vision-based tracker to get the chair bounding box cross frames.
With the above information, you can monitor the status of the same chair. like empty → occupied or occupied → empty.
Thank you for the reply! That’s what I thought we would need to do. I have one more question though, probably super dumb.
How do we change the bounding boxes when detected to empty, like instead of having a full color box how do we just draw a rectangle around it or whatever shape the box is instead of just having a solid color bounding box? I’ve tried looking this up and saw similar questions on dusty’s github but no specific guidelines or ideas as for live streaming, the answers aren’t very clear and dont understand if they are doing it on live stream detection or just trying to do it on a video file? Thank you again!