An image processing apparatus includes a first detecting unit configured to detect an object based on an upper body of a person and a second detecting unit configured to detect an object based on a face of a person. The image processing apparatus determines a level of congestion of objects contained in an input image, selects the first detecting unit when the level of congestion is low, and selects the second detecting unit when the level of congestion is high. The image processing apparatus counts the number of objects detected by the selected first or second detecting unit from the image. Thus, the image processing apparatus can detect an object and count the number of objects with high precision even when the level of congestion is high and the objects tend to overlap one another.