DSGN: Deep Stereo Geometry Network for 3D Object Detection

Chen, Yilun; Liu, Shu; Shen, Xiaoyong; Jia, Jiaya

doi:10.1109/cvpr42600.2020.01255

Cited by 157 publications

(154 citation statements)

References 32 publications

Supporting

Mentioning

152

Contrasting

Order By: Relevance

“…Pseudo-LiDAR++ is taken as a baseline and therefore in the same way, Baseline * and Baseline * * denote respectively Pseudo-LiDAR++ (SDN) and Pseudo-LiDAR++ (SDN+GDC). The proposed approach is also compared with six other outstanding methods: Pseudo LiDAR [6], TLNet [8], Stereo-RCNN [7], DSGN [9], CG-Stereo [10] which are based on low-cost sensors and the original 3D object detector PointRCNN [3] which uses 64-beam LiDAR. Firstly, comparing Ours * with the Baseline * , the proposed method outperforms the baseline over all indicators, thereby showing the ef-fect of 4-beam LiDAR in the proposed stereo matching model.…”

Section: Resultsmentioning

confidence: 99%

Sparse LiDAR and Stereo Fusion (SLS-Fusion) for Depth Estimation and 3D Object Detection

Minh¹,

Khoudour²,

Crouzil³

et al. 2021

11th International Conference of Pattern Recognition Systems (ICPRS 2021)

View full text Add to dashboard Cite

The ability to accurately detect and localize objects is recognized as being the most important for the perception of selfdriving cars. From 2D to 3D object detection, the most difficult is to determine the distance from the ego-vehicle to objects. Expensive technology like LiDAR can provide a precise and accurate depth information, so most studies have tended to focus on this sensor showing a performance gap between LiDAR-based methods and camera-based methods. Although many authors have investigated how to fuse LiDAR with RGB cameras, as far as we know there are no studies to fuse LiDAR and stereo in a deep neural network for the 3D object detection task. This paper presents SLS-Fusion, a new approach to fuse data from 4-beam LiDAR and a stereo camera via a neural network for depth estimation to achieve better dense depth maps and thereby improves 3D object detection performance. Since 4-beam LiDAR is cheaper than the well-known 64-beam LiDAR, this approach is also classified as a low-cost sensorsbased method. Through evaluation on the KITTI benchmark, it is shown that the proposed method significantly improves depth estimation performance compared to a baseline method. Also when applying it to 3D object detection, a new state of the art on low-cost sensor based method is achieved.

show abstract

Section: Resultsmentioning

confidence: 99%

Sparse LiDAR and Stereo Fusion (SLS-Fusion) for Depth Estimation and 3D Object Detection

Minh¹,

Khoudour²,

Crouzil³

et al. 2021

11th International Conference of Pattern Recognition Systems (ICPRS 2021)

View full text Add to dashboard Cite

show abstract

“…As a result, they are usually not comparable to LiDAR-based methods in terms of accuracy and efficiency. Different from traditional approaches, Stereo R-CNN [16] and DSGN [17] are the two leading methods in this area. The network of Stereo R-CNN consists of a Region Proposal Network (RPN) and a regression part.…”

Section: Related Workmentioning

confidence: 99%

“…By linking these two attack impacts together, we manage to obtain evaluation results that help answer the questions raised in Section I. The models of Stereo R-CNN [16] and DSGN [17] are pretrained with 3712 data points from the KITTI object detection dataset [32]. For each experiment setting, we test 600 real driving scenarios.…”

Section: A Common Setupmentioning

confidence: 99%

See 1 more Smart Citation

Evaluating Adversarial Attacks on Driving Safety in Vision-Based Autonomous Vehicles

Zhang

Lou

Wang

et al. 2022

IEEE Internet Things J.

View full text Add to dashboard Cite

In recent years, many deep learning models have been adopted in autonomous driving. At the same time, these models introduce new vulnerabilities that may compromise the safety of autonomous vehicles. Specifically, recent studies have demonstrated that adversarial attacks can cause a significant decline in detection precision of deep learning-based 3D object detection models. Although driving safety is the ultimate concern for autonomous driving, there is no comprehensive study on the linkage between the performance of deep learning models and the driving safety of autonomous vehicles under adversarial attacks. In this paper, we investigate the impact of two primary types of adversarial attacks, perturbation attacks and patch attacks, on the driving safety of vision-based autonomous vehicles rather than the detection precision of deep learning models. In particular, we consider two state-of-the-art models in visionbased 3D object detection, Stereo R-CNN and DSGN. To evaluate driving safety, we propose an end-to-end evaluation framework with a set of driving safety performance metrics. By analyzing the results of our extensive evaluation experiments, we find that (1) the attack's impact on the driving safety of autonomous vehicles and the attack's impact on the precision of 3D object detectors are decoupled, and (2) the DSGN model demonstrates stronger robustness to adversarial attacks than the Stereo R-CNN model. In addition, we further investigate the causes behind the two findings with an ablation study. The findings of this paper provide a new perspective to evaluate adversarial attacks and guide the selection of deep learning models in autonomous driving.

show abstract

“…The former approach always requires two or more subnetworks and the latter approach relies heavily on 2D detection, which cannot make full use of 3D geometric information. Recently, Chen et al [ 99 ] established an end-to-end method to estimate depth and detect 3D objects jointly. They encoded 3D geometry and semantic information by transforming a plane-sweep volume into a 3D geometric volume that bridges the gap between 2D images and 3D space.…”

Section: Using High-level Features In Dynamic Slammentioning

confidence: 99%

A survey: which features are required for dynamic visual simultaneous localization and mapping?

Zheng

2021

Vis. Comput. Ind. Biomed. Art

View full text Add to dashboard Cite

In recent years, simultaneous localization and mapping in dynamic environments (dynamic SLAM) has attracted significant attention from both academia and industry. Some pioneering work on this technique has expanded the potential of robotic applications. Compared to standard SLAM under the static world assumption, dynamic SLAM divides features into static and dynamic categories and leverages each type of feature properly. Therefore, dynamic SLAM can provide more robust localization for intelligent robots that operate in complex dynamic environments. Additionally, to meet the demands of some high-level tasks, dynamic SLAM can be integrated with multiple object tracking. This article presents a survey on dynamic SLAM from the perspective of feature choices. A discussion of the advantages and disadvantages of different visual features is provided in this article.

show abstract

DSGN: Deep Stereo Geometry Network for 3D Object Detection

Cited by 157 publications

References 32 publications

Sparse LiDAR and Stereo Fusion (SLS-Fusion) for Depth Estimation and 3D Object Detection

Sparse LiDAR and Stereo Fusion (SLS-Fusion) for Depth Estimation and 3D Object Detection

Evaluating Adversarial Attacks on Driving Safety in Vision-Based Autonomous Vehicles

A survey: which features are required for dynamic visual simultaneous localization and mapping?

Contact Info

Product

Resources

About