Viewpoint estimation for objects with convolutional neural network trained on synthetic images

Yumeng Wang, Shuyang Li, Mengyao Jia, Wei Liang*

*此作品的通讯作者

科研成果: 书/报告/会议事项章节会议稿件同行评审

9 引用 (Scopus)

摘要

In this paper, we propose a method to estimate object viewpoint from a single RGB image and address two problems in estimation: generating training data with viewpoint annotations and extracting powerful features for the estimation. We first collect 1780 high quality 3D CAD object models of 3 categories. Then we generate a synthetic RGB image dataset with viewpoint annotations, in which each image is generated by placing one model in a realistic panorama scene and rendering the model with a random camera parameters. We train a CNN model on our synthetic dataset to predict the object viewpoint. The proposed method is evaluated on PASCAL 3D+ dataset and our synthetic dataset. The experiment results show good performance.

源语言英语
主期刊名Advances in Multimedia Information Processing – 17th Pacific-Rim Conference on Multimedia, PCM 2016, Proceedings
编辑Enqing Chen, Yun Tie, Yihong Gong
出版商Springer Verlag
169-179
页数11
ISBN(印刷版)9783319488950
DOI
出版状态已出版 - 2016
活动17th Pacific-Rim Conference on Multimedia, PCM 2016 - Xi’an, 中国
期限: 15 9月 201616 9月 2016

出版系列

姓名Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
9917 LNCS
ISSN(印刷版)0302-9743
ISSN(电子版)1611-3349

会议

会议17th Pacific-Rim Conference on Multimedia, PCM 2016
国家/地区中国
Xi’an
时期15/09/1616/09/16

指纹

探究 'Viewpoint estimation for objects with convolutional neural network trained on synthetic images' 的科研主题。它们共同构成独一无二的指纹。

引用此