This report presents a level set approach for recovering shapes of objects from an image. Level set methods were first introduced by Osher and Sethian [4]. They showed how to propagate interfaces under various constraints on the speed to obtain an entropy-satisfying solution. Most of the ideas originated from the equations used in fluid mechanics and gas laws. Malladi,Sethian and Vemuri [3] later showed how to use these ideas for shape modeling in images. The results obtained by implementing the technique described in [3] are shown in this report and a detailed description of their technique is also presented. The primary advantage of this technique is that it can model complex shape features such as sharp corners. In fact, the contour can split and merge to detect multiple objects in the image. The equations are also easy to extend to higher dimensions.