@inproceedings{d2577013658c4af9b6b88407f1f09040,
title = "Viewpoint estimation for objects with convolutional neural network trained on synthetic images",
abstract = "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.",
keywords = "Convolutional neural network, Panorama scene rendering, Synthetic image, Viewpoint estimation",
author = "Yumeng Wang and Shuyang Li and Mengyao Jia and Wei Liang",
note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG 2016.; 17th Pacific-Rim Conference on Multimedia, PCM 2016 ; Conference date: 15-09-2016 Through 16-09-2016",
year = "2016",
doi = "10.1007/978-3-319-48896-7_17",
language = "English",
isbn = "9783319488950",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "169--179",
editor = "Enqing Chen and Yun Tie and Yihong Gong",
booktitle = "Advances in Multimedia Information Processing – 17th Pacific-Rim Conference on Multimedia, PCM 2016, Proceedings",
address = "Germany",
}