Intruder detection systems are common place now, and readily available in industry, but how do they work? They must detect people and large animals, but not generate false alarms in the presence of small animals, changes in lighting, environmental motion such as trees, or melting snow. To work correctly, the system must learn the background, in order to differentiate foreground objects. There are 2 datasets - one of an intruder and one of deer. Watershed Algorithm and MOG has been used for intruder detection.