Abstract: This article intends to propose a technique for categorizing 3D deformable objects from unprocessed single-view images. The proposed technique is built on an autoencoder, which considers the ...