ABO: Dataset and Benchmarks for Real-World 3D Object Understanding

Collins, Jasmine; Goel, Shubham; Deng, Kenan; Luthra, Achleshwar; Xu, Leon L.; Gundogdu, Erhan; Zhang, Xi; Vicente, Tomás F. Yago; Dideriksen, T.L.; Arora, Himanshu; Guillaumin, Matthieu; Malik, Jitendra

doi:10.1109/cvpr52688.2022.02045

Cited by 76 publications

(36 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We show results from images of 6 real-world objects from the ABO dataset (CC-BY-NC 4.0) [5]. Pixel-thresholding the white-background images gives masks.…”

Section: Results On Amazon Productsmentioning

confidence: 99%

“…To stress test our approach without relying on 3D point Figure 6. DS evaluated on real-world product images from Amazon [5]. For each example, we show input views (left) and reconstructed shape and texture for novel views (right).…”

Section: Results On Tanks and Templesmentioning

confidence: 99%

“…We test our differentiable stereopsis approach, which we call DS, on three datasets: Google's Scanned Objects [42], Tanks and Temples [24] and the Amazon-Berkeley Objects [5]. We additionally evaluate on DTU MVS [18] in the Appendix.…”

Section: Methodsmentioning

confidence: 99%

“…For example, Amazon, eBay or Facebook Marketplace only contain a handful of views for each listed item, and any 3D reconstruction has to originate from 10 views or less. We emphasize on this harder, yet more realistic, setting and show empirical results with real product images from Amazon [5]. On Google's Scanned Objects [42] we perform an extensive quantitative and qualitative analysis and compare to competing approaches under settings similar to ours.…”

Section: Algorithm 1: Differentiable Stereopsis (2-view)mentioning

confidence: 99%

See 3 more Smart Citations

Differentiable Stereopsis: Meshes from multiple views using differentiable rendering

Goel

Gkioxari²,

Malik

2022

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

show abstract

“…We show results from images of 6 real-world objects from the ABO dataset (CC-BY-NC 4.0) [5]. Pixel-thresholding the white-background images gives masks.…”

Section: Results On Amazon Productsmentioning

confidence: 99%

Section: Results On Tanks and Templesmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Algorithm 1: Differentiable Stereopsis (2-view)mentioning

confidence: 99%

See 2 more Smart Citations

Differentiable Stereopsis: Meshes from multiple views using differentiable rendering

Goel

Gkioxari²,

Malik

2022

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

show abstract

“…Compared with 2D representations such as images, 3D data generally offers a more comprehensive understanding for the real world [8], [39], [21], [5]. Especially, with the rapid development of 3D sensors (e.g., LiDAR), dealing with the easily accessible 3D data has received growing attention, which necessitates AD techniques to provide quality inspection in various manufacturing scenarios.…”

Section: Introductionmentioning

confidence: 99%

Teacher-Student Network for 3D Point Cloud Anomaly Detection with Few Normal Samples

Qin¹,

Gu²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Anomaly detection, which is a critical and popular topic in computer vision, aims to detect anomalous samples that are different from the normal (i.e., non-anomalous) ones. The current mainstream methods focus on anomaly detection for images, whereas little attention has been paid to 3D point cloud. In this paper, drawing inspiration from the knowledge transfer ability of teacher-student architecture and the impressive feature extraction capability of recent neural networks, we design a teacher-student structured model for 3D anomaly detection. Specifically, we use feature space alignment, dimension zoom, and max pooling to extract the features of the point cloud and then minimize a multi-scale loss between the feature vectors produced by the teacher and the student networks. Moreover, our method only requires very few normal samples to train the student network due to the teacher-student distillation mechanism. Once trained, the teacher-student network pair can be leveraged jointly to fulfill 3D point cloud anomaly detection based on the calculated anomaly score. For evaluation, we compare our method against the reconstruction-based method on the ShapeNet-Part dataset. The experimental results and ablation studies quantitatively and qualitatively confirm that our model can achieve higher performance compared with the state of the arts in 3D anomaly detection with very few training samples.

show abstract

TO-Scene: A Large-Scale Dataset for Understanding 3D Tabletop Scenes

Chen

Liu

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

ABO: Dataset and Benchmarks for Real-World 3D Object Understanding

Cited by 76 publications

References 32 publications

Differentiable Stereopsis: Meshes from multiple views using differentiable rendering

Differentiable Stereopsis: Meshes from multiple views using differentiable rendering

Teacher-Student Network for 3D Point Cloud Anomaly Detection with Few Normal Samples

TO-Scene: A Large-Scale Dataset for Understanding 3D Tabletop Scenes

Contact Info

Product

Resources

About