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