Snagging Parking Spaces with Mask R-CNN and Python – Adam Geitgey – Medium

We'll pass each frame of video through the pipeline, one frame at a time. The first step in the pipeline is to detect all possible parking spaces in a frame of video. Obviously we need to know which parts of the image are parking spaces before we can detect which parking spaces are unoccupied. The second step is to detect all the cars in each frame of video. This will let us track the movement of each car from frame to frame.