Training and Testing Object Detectors With Virtual Images

作     者：Yonglin Tian Xuan Li Kunfeng Wang Fei-Yue Wang 

作者机构：Department of Automation University of Science and Technology of China Hefei 230027 China State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijirtg 100190 China School of Automation Beijing Institute of Technology Beijing 100081. China Qingdao Academy of Intelligent Industries Qingdao 266000 China Research Center for Computational Experiments and Parallel Systems Technology National University of Defense Technology Changsha 410073 China IEEE 

出 版 物：《IEEE/CAA Journal of Automatica Sinica》 (自动化学报（英文版）)

年 卷 期：2018年第5卷第2期

页      面：539-546页

核心收录：

学科分类：0810[工学-信息与通信工程] 1205[管理学-图书情报与档案管理] 08[工学] 0802[工学-机械工程] 081101[工学-控制理论与控制工程] 0811[工学-控制科学与工程] 081102[工学-检测技术与自动化装置] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：supported by the National Natural Science Foundation of China(61533019 71232006) 

主　　题：Deep learning object detection parallel vision virtual dataset 

摘      要：In the area of computer vision, deep learning has produced a variety of state-of-the-art models that rely on massive labeled data. However, collecting and annotating images from the real world is too demanding in terms of labor and money investments, and is usually inflexible to build datasets with specific characteristics, such as small area of objects and high occlusion level. Under the framework of Parallel Vision, this paper presents a purposeful way to design artificial scenes and automatically generate virtual images with precise annotations.A virtual dataset named Parallel Eye is built, which can be used for several computer vision tasks. Then, by training the DPM(Deformable parts model) and Faster R-CNN detectors, we prove that the performance of models can be significantly improved by combining Parallel Eye with publicly available real-world datasets during the training phase. In addition, we investigate the potential of testing the trained models from a specific aspect using intentionally designed virtual datasets, in order to discover the flaws of trained models. From the experimental results, we conclude that our virtual dataset is viable to train and test the object detectors.