I am looking for state-of-the-art algorithms which are being followed for the detection of events in videos given only a single camera footage and also considering the camera is not calibrated i.e. no information can be deduced which can help us in finding the 3D locations of the objects present in the scene?