In clinical diagnosis, segmentation of cardiac magnetic resonance imaging (MRI) plays a major role. For distinguishing inner and outer layer borders in the left and right ventricles of the human heart, automated segmentation methods are used instead of traditional diagnostic process which is very slow and requires more labors. This paper discusses some of the main segmentation methods such as image based, pixel classification and deformable models which are based on weak or no prior which means it does not require any prior knowledge to understand. Among these, multilevel thresholding, fully convolutional neural network (FCN) and active contour-based segmentation methods are mainly focused. These methods are able to successfully segment left ventricle (LV) and right ventricle (RV) and are able to achieve better performance in the classification of cardiac diseases.