Shape Prior Deformation for Categorical 6D Object Pose and Size Estimation

Tian, Meng; Ang, Marcelo H.; Lee, Gim Hee

doi:10.1007/978-3-030-58589-1_32

Cited by 117 publications

(217 citation statements)

References 40 publications

Supporting

Mentioning

216

Contrasting

Order By: Relevance

“…We computed the Chamfer Distance of the reconstructed point cloud with the ground truth point cloud and compared it with other reconstruction types used by other methods. From Table 2, we can see that the average reconstruction error of our method is 0.86, which is 72.9% and 18.9% lower than that of Shape-Prior [34] and CASS [4], respectively. It shows that our method achieves better pose estimation results via a simpler reconstruction task.…”

Section: Evaluation Of Reconstructionmentioning

confidence: 89%

“…The 6D pose and size are calculated by the Umeyama algorithm [36] with the NOCS map and the observed points. Shape-Prior [34] adopted similar method with [40], but both extra shape prior knowledge and dense-fusion feature [39], instead of RGB feature, are used. CASS [4] estimated the 6D pose via the learning of a canonical shape space with dense-fusion feature [39].…”

Section: Category-level Pose Estimationmentioning

confidence: 99%

“…Without loss of generality, we assume that the green vector is along the symmetry axis; then, we set λ r as zero to handle the circular symmetry objects. For other types of symmetric objects, we can employ the rotation mapping function used in [24,34] to map the relevant rotation matrices to a unique one.…”

Section: Decoupled Rotation Estimationmentioning

confidence: 99%

“…For category-level pose estimation, we adopt the same metrics, 3D IoU 25,50,75 and n • m cm used in [40,4,34]. For instance-level pose estimation, we compare the performance of FS-Net with other state-of-the-art instance-level methods using the ADD-(S) metric [12].…”

Section: Evaluation Metricsmentioning

confidence: 99%

“…Chen et al [4] proposed to learn a canonical shape space (CASS) to tackle intra-class shape variations with RGB-D fusion feature [39]. Tian et al [34] trained a network to predict the NOCS map of different objects, with the uniform shape prior learned from a shape collection, and RGB-D fusion feature [39]. Although these methods achieved state-of-the-art performance, there are still two remaining issues.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

FS-Net: Fast Shape-based Network for Category-Level 6D Object Pose Estimation with Decoupled Rotation Mechanism

Chen

Jia

Chang

et al. 2021

2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

110

View full text Add to dashboard Cite

In this paper, we focus on category-level 6D pose and size estimation from a monocular RGB-D image. Previous methods suffer from inefficient category-level pose feature extraction, which leads to low accuracy and inference speed. To tackle this problem, we propose a fast shapebased network (FS-Net) with efficient category-level feature extraction for 6D pose estimation. First, we design an orientation aware autoencoder with 3D graph convolution for latent feature extraction. Thanks to the shift and scaleinvariance properties of 3D graph convolution, the learned latent feature is insensitive to point shift and object size. Then, to efficiently decode category-level rotation information from the latent feature, we propose a novel decoupled rotation mechanism that employs two decoders to complementarily access the rotation information. For translation and size, we estimate them by two residuals: the difference between the mean of object points and ground truth translation, and the difference between the mean size of the category and ground truth size, respectively. Finally, to increase the generalization ability of the FS-Net, we propose an online box-cage based 3D deformation mechanism to augment the training data. Extensive experiments on two benchmark datasets show that the proposed method achieves state-ofthe-art performance in both category-and instance-level 6D object pose estimation. Especially in category-level pose estimation, without extra synthetic data, our method outperforms existing methods by 6.3% on the NOCS-REAL dataset 1 .

show abstract

Section: Evaluation Of Reconstructionmentioning

confidence: 89%