• Corpus ID: 237091793

SPG: Unsupervised Domain Adaptation for 3D Object Detection via Semantic Point Generation

@article{Xu2021SPGUD,
  title={SPG: Unsupervised Domain Adaptation for 3D Object Detection via Semantic Point Generation},
  author={Qiangeng Xu and Yin Zhou and Weiyue Wang and C. Qi and Drago Anguelov},
  journal={ArXiv},
  year={2021},
  volume={abs/2108.06709}
}
In autonomous driving, a LiDAR-based object detector should perform reliably at different geographic locations and under various weather conditions. While recent 3D detection research focuses on improving performance within a single domain, our study reveals that the performance of modern detectors can drop drastically cross-domain. In this paper, we investigate unsupervised domain adaptation (UDA) for LiDAR-based 3D object detection. On the Waymo Domain Adaptation dataset, we identify the… 
See Eye to Eye: A Lidar-Agnostic 3D Detection Framework for Unsupervised Multi-Target Domain Adaptation
TLDR
This work proposes a novel unsupervised multi-target domain adaptation framework, SEE, for transferring the performance of state-of-the-art 3D detectors across both fixed and flexible scan pattern lidars without requiring fine-tuning of models by end-users.
Behind the Curtain: Learning Occluded Shapes for 3D Object Detection
  • Qiangeng Xu, Yiqi Zhong, U. Neumann
  • Computer Science
    ArXiv
  • 2021
TLDR
A novel LiDAR-based 3D object detection model, dubbed Behind the Curtain Detector (BtcDet), which learns the object shape priors and estimates the complete object shapes that are partially occluded (curtained) in point clouds.
3D Object Detection Combining Semantic and Geometric Features from Point Clouds
TLDR
The VTPM is a Voxel-Point-Based Module that finally implements 3D object detection in point space, which is more conducive to the detection of small-size objects and avoids the presets of anchors in inference stage.
Investigating the Impact of Multi-LiDAR Placement on Object Detection for Autonomous Driving
TLDR
This study introduces an easy-to-compute information-theoretic surrogate metric and shows that sensor placement is non-negligible in 3D point cloud-based object detection, which will contribute to 5% ∼ 10% performance discrepancy in terms of average precision in challenging 3D object detection settings.
DSOR: A Scalable Statistical Filter for Removing Falling Snow from LiDAR Point Clouds in Severe Winter Weather
TLDR
The Winter Adverse Driving dataSet (WADS) collected in the snow belt region of Michigan’s Upper Peninsula is introduced and the Dynamic Statistical Outlier Removal (DSOR) filter is presented, a statistical PCL-based filter capable or removing snow with a higher recall than the state of the art snow de-noising filter while being 28% faster.
The Norm Must Go On: Dynamic Unsupervised Domain Adaptation by Normalization
TLDR
This work modify the feature representations of the model by continuously adapting the statistics of the batch normalization layers, and shows that by accessing only a tiny fraction of unlabeled data from the shifted domain and adapting sequentially, a strong performance gain can be achieved.

References

SHOWING 1-10 OF 84 REFERENCES
Associate-3Ddet: Perceptual-to-Conceptual Association for 3D Point Cloud Object Detection
  • Liang Du, Xiaoqing Ye, +4 authors Shilei Wen
  • Computer Science, Engineering
    2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2020
TLDR
A domain adaptation like approach is innovatively proposed to enhance the robustness of the feature representation in 3D object detection and mimics the functionality of the human brain when proceeding object perception.
SqueezeSegV2: Improved Model Structure and Unsupervised Domain Adaptation for Road-Object Segmentation from a LiDAR Point Cloud
TLDR
This work introduces a new model SqueezeSegV2, which is more robust against dropout noises in LiDAR point cloud and therefore achieves significant accuracy improvement, and a domain-adaptation training pipeline consisting of three major components: learned intensity rendering, geodesic correlation alignment, and progressive domain calibration.
Unsupervised scene adaptation for semantic segmentation of urban mobile laser scanning point clouds
TLDR
A novel method to perform unsupervised scene adaptation for semantic segmentation of urban MLS point clouds by proposing a new pointwise attentive transformation module (PW-ATM) to adaptively perform the data alignment and an end-to-end alignment deep network architecture is designed.
LaserNet: An Efficient Probabilistic 3D Object Detector for Autonomous Driving
TLDR
benchmark results show that this approach has significantly lower runtime than other recent detectors and that it achieves state-of-the-art performance when compared on a large dataset that has enough data to overcome the challenges of training on the range view.
SARPNET: Shape attention regional proposal network for liDAR-based 3D object detection
TLDR
A novel 3D object detection network called SARPNET is introduced, which deploys a new low-level feature encoder to remedy the sparsity and inhomogeneity of LiDAR point clouds with an even sample method, and embodies a shape attention mechanism that learns the statistic 3D shape priors of objects and uses them to spatially enhance semantic embeddings.
EPNet: Enhancing Point Features with Image Semantics for 3D Object Detection
TLDR
A novel fusion module is proposed to enhance the point features with semantic image features in a point-wise manner without any image annotations to address two critical issues in the 3D detection task, including the exploitation of multiple sensors~ and the inconsistency between the localization and classification confidence.
Structure Aware Single-Stage 3D Object Detection From Point Cloud
TLDR
An auxiliary network is designed which converts the convolutional features in the backbone network back to point-level representations and an efficient part-sensitive warping operation is developed to align the confidences to the predicted bounding boxes.
Domain Adaptation for Vehicle Detection from Bird's Eye View LiDAR Point Cloud Data
TLDR
The proposed framework is based on the deep cycle-consistent generative adversarial networks (CycleGAN) architecture and has shown competitive results with an improvement of more than 7% in average precision score over other baseline approaches when tested on real BEV point cloud images.
SegVoxelNet: Exploring Semantic Context and Depth-aware Features for 3D Vehicle Detection from Point Cloud
TLDR
A semantic context encoder is proposed to leverage the free-of-charge semantic segmentation masks in the bird's eye view to address the semantic context information in LiDAR.
End-to-End Multi-View Fusion for 3D Object Detection in LiDAR Point Clouds
TLDR
This paper aims to synergize the birds-eye view and the perspective view and proposes a novel end-to-end multi-view fusion (MVF) algorithm, which can effectively learn to utilize the complementary information from both and significantly improves detection accuracy over the comparable single-view PointPillars baseline.
...
1
2
3
4
5
...