This is what has been on my mind when seeing the video. It was odd, to me, the way he covered the camera lens. You would think you would just use your palm or fingers. However, if he was in mid knock then the strange way he blocked does make sense. If you watch the video it really does seem he was going for the door, looking to the left of it first (looking through the sidelight?). He then raises his hand as if to knock (actually pound) on the door then seems to notice the camera. He actually steps to his right. After covering the camera you can see his left hand do something. At first his left hand is wide as if to grab the doorbell then changes his mind and it really looks like he presses the doorbell. That doesn't make sense. Its between eight and ten seconds of the video released. His left hand presses on something below the lens. The doorknob is clearly seen and not what he is touching.
For some reason after manipulating the doorbell he decides to turn around and find the plant etc.. Unfortunately we don't have video if the camera comes off while the he is holding the plant over it or something else happens. Not having sound also leaves a gap in possibly knowing what is going on. Of course the intention of the released portion of the video was possible recognition of the person and not what he is doing.