We propose a method to recognize and classify inverse synthetic-aperture radar (ISAR) images of a target. The information that is combined from various image frames, it is generally in the context of time-averaging to remove statistically atomic noise shifts in the images. Due to wave action, a ship has constantly changing roll, yaw and pitch angular velocities, which makes the ISAR images quite changeable from frame to frame. A method for identifying the target based on 3D dispersed information from a sequence of 2D ISAR images is elucidated. A Trained-Model will be given an ISAR image as an input; and this model will use an image classifier based on deep learning to recognize and classify the images.