This paper investigates the benefits of incorporating point visibility information of 3D point clouds within a deep learning framework, using occlusion-aware 3D priors. The presented methods for deriving the visibility of each point rely on ray-casting techniques, making the proposed solution generic and sensor independent. We demonstrate the benefits of integrating point visibility using two real-world applications. In a first application, a novel data augmentation technique is proposed leveraging occlusion-aware CAD 3D priors, resulting in state-of-the-art 3D vehicle detection. In a second application, we integrate visibility information into a vehicle pose estimation pipeline based on 3D priors. The presented techniques achieve state-of-the-art performance, significantly improving both translation and rotation accuracy on the Apollo3DCar dataset.
Ducastel, O, Yangxintong, L, Denis, L & Munteanu, A 2023, Occlusion-Aware 3D Priors for Deep Learning-Based Applications. in 2023 IEEE 25th International Workshop on Multimedia Signal Processing, MMSP 2023. 2023 IEEE 25th International Workshop on Multimedia Signal Processing, MMSP 2023, IEEE, pp. 1-6, IEEE 25th International Workshop on Multimedia Signal Processing, Poitiers, France, 27/09/23. https://doi.org/10.1109/MMSP59012.2023.10337644
Ducastel, O., Yangxintong, L., Denis, L., & Munteanu, A. (2023). Occlusion-Aware 3D Priors for Deep Learning-Based Applications. In 2023 IEEE 25th International Workshop on Multimedia Signal Processing, MMSP 2023 (pp. 1-6). (2023 IEEE 25th International Workshop on Multimedia Signal Processing, MMSP 2023). IEEE. https://doi.org/10.1109/MMSP59012.2023.10337644
@inproceedings{c0355019395f474fb07d3a883bf8cab3,
title = "Occlusion-Aware 3D Priors for Deep Learning-Based Applications",
abstract = "This paper investigates the benefits of incorporating point visibility information of 3D point clouds within a deep learning framework, using occlusion-aware 3D priors. The presented methods for deriving the visibility of each point rely on ray-casting techniques, making the proposed solution generic and sensor independent. We demonstrate the benefits of integrating point visibility using two real-world applications. In a first application, a novel data augmentation technique is proposed leveraging occlusion-aware CAD 3D priors, resulting in state-of-the-art 3D vehicle detection. In a second application, we integrate visibility information into a vehicle pose estimation pipeline based on 3D priors. The presented techniques achieve state-of-the-art performance, significantly improving both translation and rotation accuracy on the Apollo3DCar dataset.",
author = "Olivier Ducastel and Lyu Yangxintong and Leon Denis and Adrian Munteanu",
note = "Funding Information: The authors would like to thank Innoviris (TORRES, SPECTRE) for their financial support. Publisher Copyright: {\textcopyright} 2023 IEEE.; IEEE 25th International Workshop on Multimedia Signal Processing, IEEE MMSP 2023 ; Conference date: 27-09-2023 Through 29-09-2023",
year = "2023",
month = sep,
day = "27",
doi = "10.1109/MMSP59012.2023.10337644",
language = "English",
series = "2023 IEEE 25th International Workshop on Multimedia Signal Processing, MMSP 2023",
publisher = "IEEE",
pages = "1--6",
booktitle = "2023 IEEE 25th International Workshop on Multimedia Signal Processing, MMSP 2023",
}