12 Hits in 2.0 sec

ST3D: Self-training for Unsupervised Domain Adaptation on 3D Object Detection [article]

Jihan Yang, Shaoshuai Shi, Zhe Wang, Hongsheng Li, Xiaojuan Qi
2021 arXiv   pre-print
We present a new domain adaptive self-training pipeline, named ST3D, for unsupervised domain adaptation on 3D object detection from point clouds.  ...  First, we pre-train the 3D detector on the source domain with our proposed random object scaling strategy for mitigating the negative effects of source domain bias.  ...  Quality of pseudo labels on KITTI training set. Conclusion We have presented ST3D -a redesigned self-training pipeline -for unsupervised domain adaptive 3D object detection from point clouds.  ... 
arXiv:2103.05346v2 fatcat:kdmjr5yy5vcx3baf3joqj3rj4u

ST3D++: Denoised Self-training for Unsupervised Domain Adaptation on 3D Object Detection [article]

Jihan Yang, Shaoshuai Shi, Zhe Wang, Hongsheng Li, Xiaojuan Qi
2021 arXiv   pre-print
In this paper, we present a self-training method, named ST3D++, with a holistic pseudo label denoising pipeline for unsupervised domain adaptation on 3D object detection.  ...  First, ST3D++ pre-trains the 3D object detector on the labeled source domain with random object scaling (ROS) which is designed to reduce target domain pseudo label noise arising from object scale bias  ...  the effectiveness of self-training on domain adaptive 3D object detection.  ... 
arXiv:2108.06682v1 fatcat:nhpe3pvcufeabhtme5ou2fvxxi

Uncertainty-aware Mean Teacher for Source-free Unsupervised Domain Adaptive 3D Object Detection [article]

Deepti Hegde, Vishwanath Sindagi, Velat Kilic, A. Brinton Cooper, Mark Foster, Vishal Patel
2021 arXiv   pre-print
Pseudo-label based self training approaches are a popular method for source-free unsupervised domain adaptation.  ...  We demonstrate our method on several domain adaptation scenarios, from cross-dataset to cross-weather conditions, and achieve state-of-the-art performance in these cases, on the KITTI lidar target dataset  ...  We propose a framework for unsupervised domain adaptation for 3D object detection.  ... 
arXiv:2109.14651v1 fatcat:5wo4dotrb5ejtnqcgqneagnkyi

Attentive Prototypes for Source-free Unsupervised Domain Adaptive 3D Object Detection [article]

Deepti Hegde, Vishal M. Patel
2021 arXiv   pre-print
3D object detection networks tend to be biased towards the data they are trained on.  ...  We propose a single-frame approach for source-free, unsupervised domain adaptation of lidar-based 3D object detectors that uses class prototypes to mitigate the effect pseudo-label noise.  ...  Yang et al. put forth ST3D [40] , a self training approach for 3D domain adaptive object detection where the network is adapted by training with a proposed curriculum data augmentation algorithm using  ... 
arXiv:2111.15656v2 fatcat:6lzmcwufyzfxtni53swxkiuohi

Voxel-MAE: Masked Autoencoders for Pre-training Large-scale Point Clouds [article]

Chen Min and Dawei Zhao and Liang Xiao and Yiming Nie and Bin Dai
2022 arXiv   pre-print
We also validate the effectiveness of Voxel-MAE in unsupervised domain adaptative tasks, which proves the generalization ability of Voxel-MAE.  ...  Extensive experiments show great effectiveness of our pre-trained model with 3D object detectors (SECOND, CenterPoint, and PV-RCNN) on two popular datasets (KITTI, Waymo).  ...  Domain Adaptative 3D Object Detection Recently, some works that focus on addressing the domain shift on point clouds in the 3D object detection task have been proposed.  ... 
arXiv:2206.09900v3 fatcat:zpjavedxtjfylghtycce2wvrba

See Eye to Eye: A Lidar-Agnostic 3D Detection Framework for Unsupervised Multi-Target Domain Adaptation [article]

Darren Tsai and Julie Stephany Berrio and Mao Shan and Stewart Worrall and Eduardo Nebot
2021 arXiv   pre-print
This leads to performance degradation when 3D detectors trained for one lidar are tested on other types of lidars.  ...  We explicitly deal with the sampling discrepancy by proposing a novel unsupervised multi-target domain adaptation framework, SEE, for transferring the performance of state-of-the-art 3D detectors across  ...  We adopt a similar convention to [64] and compare SEE with (1) Source-only, where no DA strategies are used; (2) ST3D [64] , the SOTA UDA method on 3D object detection using self-training.  ... 
arXiv:2111.09450v1 fatcat:7ywxl7lhffdmvi7avlqq6pdb2u

One Million Scenes for Autonomous Driving: ONCE Dataset [article]

Jiageng Mao, Minzhe Niu, Chenhan Jiang, Hanxue Liang, Jingheng Chen, Xiaodan Liang, Yamin Li, Chaoqiang Ye, Wei Zhang, Zhenguo Li, Jie Yu, Hang Xu (+1 others)
2021 arXiv   pre-print
In this paper, we introduce the ONCE (One millioN sCenEs) dataset for 3D object detection in the autonomous driving scenario.  ...  To facilitate future research on exploiting unlabeled data for 3D detection, we additionally provide a benchmark in which we reproduce and evaluate a variety of self-supervised and semi-supervised methods  ...  Unsupervised Domain Adaptation for 3D Object Detection Unsupervised domain adaptation for 3D object detection aims to adapt a detection model from the source dataset to the target dataset without supervisory  ... 
arXiv:2106.11037v3 fatcat:fwgrb57yarhujmetzpewtdzzei

Real-Time and Robust 3D Object Detection Within Road-Side LiDARs Using Domain Adaptation [article]

Walter Zimmer, Marcus Grabler, Alois Knoll
2022 arXiv   pre-print
This work aims to address the challenges in domain adaptation of 3D object detection using infrastructure LiDARs.  ...  We apply domain adaptation from the semi-synthetic A9-Dataset to the semi-synthetic dataset from the Regensburg Next project by applying transfer learning and achieve a 3D mAP@0.25 of 93.49% on the Car  ...  ST3D [36] provides a self-training pipeline for unsupervised domain adaptation on 3D object detection from point clouds, where no annotated data in the target domain is available.  ... 
arXiv:2204.00132v1 fatcat:w7fxwi4vzng7dcjm6t4d63tn2m

Unsupervised Domain Adaptation for Monocular 3D Object Detection via Self-Training [article]

Zhenyu Li, Zehui Chen, Ang Li, Liangji Fang, Qinhong Jiang, Xianming Liu, Junjun Jiang
2022 arXiv   pre-print
Then, we propose STMono3D, a new self-teaching framework for unsupervised domain adaptation on Mono3D.  ...  STMono3D achieves remarkable performance on all evaluated datasets and even surpasses fully supervised results on the KITTI 3D object detection dataset.  ...  We call this task unsupervised domain adaptation (UDA) for monocular 3D object detection.  ... 
arXiv:2204.11590v2 fatcat:oho5oi4go5fxtavavfkqeoiqme

Exploiting Playbacks in Unsupervised Domain Adaptation for 3D Object Detection [article]

Yurong You, Carlos Andres Diaz-Ruiz, Yan Wang, Wei-Lun Chao, Bharath Hariharan, Mark Campbell, Kilian Q Weinberger
2022 arXiv   pre-print
Self-driving cars must detect other vehicles and pedestrians in 3D to plan safe routes and avoid collisions.  ...  State-of-the-art 3D object detectors, based on deep learning, have shown promising accuracy but are prone to over-fit to domain idiosyncrasies, making them fail in new environments -- a serious problem  ...  Concurrent work ST3D [47] also tackles UDA in 3D via self-training, but it focuses on addressing the object size discrepancy across domains [7] .  ... 
arXiv:2103.14198v2 fatcat:cwzaqs7zqnejnf2neemlhv4rae

Towards Adaptive Unknown Authentication for Universal Domain Adaptation by Classifier Paradox [article]

Yunyun Wang, Yao Liu, Songcan Chen
2022 arXiv   pre-print
Universal domain adaptation (UniDA) is a general unsupervised domain adaptation setting, which addresses both domain and label shifts in adaptation.  ...  Further, instead of feature alignment for shared classes, implicit domain alignment is conducted in output space such that samples across domains share the same decision boundary, though with feature discrepancy  ...  St3d: Self-training for unsupervised domain adaptation on 3d object detection. Proceedings of the ieee/cvf conference on computer vision and pattern recognition (pp. 10368-10378).  ... 
arXiv:2207.04494v1 fatcat:encctzywqzeihctvbo6qx7vlri


Qiyu Sun, Chaoqiang Zhao, Yang Tang, Feng Qian
2021 Scientia Sinica Technologica  
of the 2019 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019, 780-790 [61] Yang J, Shi S, Wang Z, Li H, Qi X, St3d: Selftraining for unsupervised domain adaptation on 3d objectdetection  ...  attention for category-aware unsupervised domain adaptive object detection, In: Proceedings of the 2021 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2021 [59] Kim T, Jeong M, Kim  ... 
doi:10.1360/sst-2021-0150 fatcat:tutl3s656re55dl6vixh32nkai