In this study, a method for a robot to recall multiple grasping methods for a given object is proposed. Therobot learns grasping methods using a convolutional neural network to observethe grasping activities of human without special instructions. For this setting, only one grasping motion is observed for an object at a time. By acquiring multiple grasping methods, the robot learnsby automatically clustering the observed grasping postures. In the proposed method, the grasping methods are clustered during the process of learning the grasping position. The method first recalls the grasping positions. The network for recalling the grasping position estimates the multi-channel heatmap that indicates one grasping position in each channel.The method then checks the graspability for each estimated position.Finally, it recalls the hand shapes that are based on the estimated grasping position and the object’s shape. This study describes the results of recalling multiple grasping methods and demonstrates the effectiveness of the proposed method.