This surveillance application involves using a Raspberry Pi along with a Camera to detect if a room is occupied through motion detection. If a room is occupied it will capture an image and save the stored image to Dropbox.

Configure conf.json with important variable. This keeps our code clean and allows universal use across multiple files. Variable assigned involve resolution, frames per second and thresholds.

Integrate application with Dropbox. Create application within Dropbox App Console found at www.dropbox.com/developer.

We initialise the camera and grab a reference of it. We then loop over every frame that is being processed by the camera. Each frame is converted into an array and we apply Gaussian Blur to remove unnecessary high frequency noise. The weighted mean is then calculated of previous frames along with the current frame. This allows our background to adjust to changes such as lighting conditions throughout the day. We calculate a delta from our background model - current_frame.

Thresholding the delta image followed by dilating it will allow us to then find the contours. Contours are then looped over to see if they pass the minimum area criteria eg) are they big enough to pass for a human. Bounding box is then computed and placed on feed.

If the room has been flagged as occupied then an image is uploaded to dropbox. However it makes sure that enough time has passed so consecutive images are not uploaded of same person constantly.