Viewpoint estimation for workpieces with deep transfer learning from cold to hot

Abstract

With the revival of deep neural networks, viewpoint estimation problem can be handled by the learned distinctive features. However, the scarcity and expensiveness of viewpoint annotation for the real-world industrial workpieces impede its progress of application. In this paper, we propose a deep transfer learning method for viewpoint estimation by transferring priori knowledge from labeled synthetic images to unlabeled real images. The synthetic images are rendered from 3D Computer-Aided Design (CAD) models and annotated automatically. To boost the performance of deep transfer network, we design a new two-stage training strategy called cold-to-hot training. At the cold start stage, deep networks are trained for the joint tasks of classification and knowledge transfer in the absence of labels of real images. But after it turns into the hot stage, the pseudo labels of real images are employed for controlling the distributions of input data. The satisfactory experimental results demonstrate the effectiveness of the proposed method in dealing with the viewpoint estimation problem under the scarcity of annotated real workpiece images.