The present invention relates to a method and apparatus for human detection in images taken from a surveillance camera. In one embodiment, the method comprises subtracting a current image from a baseline image to obtain a delta image, said delta image comprising at least one object; filtering stray noises and/or noise clusters from said delta image; marking and traversing the boundary of said at least one object; determining the at least one object as valid object: if a length traversed along the boundary of said at least one object is greater than a pre-determined threshold value; and if an interior region of said at least one object is detected as being solid; erasing the interior region of said at least one object and determining one or more pre-determined points on the boundary of said valid object; determining one or more pre-determined parameters corresponding to said pre-determined points and ratio thereof; processing the ratios to identify if the object is a human being or not.