Domain Adaptation for Vehicle Detection from Bird's Eye View LiDAR Point Cloud Data

Saleh, Khaled J.; Abobakr, Ahmed; Attia, Mohamed; Iskander, Julie; Nahavandi, Darius; Hossny, Mohammed; Nahvandi, Saeid

doi:10.1109/iccvw.2019.00404

Cited by 49 publications

(28 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, upsampling the entire point cloud will lead to a significantly higher latency. A third approach is to leverage style transfer techniques: [80,40,12,20,48,21,47] render point clouds as 2D pseudo images and enforce the renderings from different domains to be resemblant in style. However, these methods introduce an information bottleneck during rasterization [79] and they are not applicable to modern pointbased 3D detectors [49].…”

Section: Previous Methods To Address the Domain Gapmentioning

confidence: 99%

See 1 more Smart Citation

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

Xu¹,

Zhou²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

Section: Previous Methods To Address the Domain Gapmentioning

confidence: 99%

“…[77,45] align the global and local features for object-level tasks. To reduce the sparsity, [59] projects the point cloud to 2D view, while [47] projects the point cloud to birds-eye view (BEV). [15] creates a car model set and adapts their features to the detection object features.…”

Section: Unsupervised Domain Adaptation For 2d Detectionmentioning

confidence: 99%

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

Xu¹,

Zhou²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Apart from UDA on object point clouds, several methods are proposed to address specific domain gaps on LiDAR point clouds, where the common factors are depth missing and sampling difference between sensors. Both [33] and [23] use CycleGAN [34] to generate more realistic Li-DAR point clouds from synthetic data, i.e., sim2real. Complete & Label [31] leverages segmentation on completed surface reconstructed from sparse point cloud for better adaptation.…”

Section: D and 3d Unsupervised Domain Adaptationmentioning

confidence: 99%

Domain Adaptation on Point Clouds via Geometry-Aware Implicits

Shen¹,

Yang²,

Yan³

et al. 2021

Preprint

View full text Add to dashboard Cite

As a popular geometric representation, point clouds have attracted much attention in 3D vision, leading to many applications in autonomous driving and robotics. One important yet unsolved issue for learning on point cloud is that point clouds of the same object can have significant geometric variations if generated using different procedures or captured using different sensors. These inconsistencies induce domain gaps such that neural networks trained on one domain may fail to generalize on others. A typical technique to reduce the domain gap is to perform adversarial training so that point clouds in the feature space can align. However, adversarial training is easy to fall into degenerated local minima, resulting in negative adaptation gains.Here we propose a simple yet effective method for unsupervised domain adaptation on point clouds by employing a self-supervised task of learning geometry-aware implicits, which plays two critical roles in one shot. First, the geometric information in the point clouds is preserved through the implicit representations for downstream tasks. More importantly, the domain-specific variations can be effectively learned away in the implicit space. We also propose an adaptive strategy to compute unsigned distance fields for arbitrary point clouds due to the lack of shape models in practice. When combined with a task loss, the proposed outperforms state-of-the-art unsupervised domain adaptation methods that rely on adversarial domain alignment and more complicated self-supervised tasks. Our method is evaluated on both PointDA-10 and GraspNet datasets. The code and trained models will be publicly available.

show abstract

“…A more recent LiDAR-focused domain adaptation survey [12] classifies methods into: 1) Domain-invariant data representation methods [13], [14], mainly based on hand-crafted data preprocessing to move different domains into a common representation (e.g. LiDAR data rotation and normalization), 2) Domain-invariant feature learning for finding a common representation space for the source and target domains [15], [16], 3) Normalization statistics that attempt to align the domain distributions by a normalization of the mean and variance of activations, and 4) Domain mapping, where source data is transformed, usually using GANs or adversarial training to appear like target data [17], [18], [19].…”

Section: Introduction and Prior Workmentioning

confidence: 99%

“…A few examples of domain mapping methods in the context of training data generation from simulated synthetic frames are [17], [18], [20], where a CycleGAN [21] is trained using unpaired simulated and real projected bird's eye view (BEV) images to generate pseudo-labeled simulated data for off-line training of a BEV YOLOv3 [22] object detection network. SqueezeSegV2 [23] is another example that uses simulators to generate large quantities of labeled spherical projections of synthetic LiDAR data to train perception models.…”

Section: Introduction and Prior Workmentioning

confidence: 99%

Domain Adaptation in LiDAR Semantic Segmentation via Alternating Skip Connections and Hybrid Learning

Corral-Soto¹,

Rochan²,

Y.³

et al. 2022

Preprint

View full text Add to dashboard Cite

In this paper we address the problem of training a LiDAR semantic segmentation network using a fully-labeled source dataset and a target dataset that only has a small number of labels. To this end, we develop a novel imageto-image translation engine, and couple it with a LiDAR semantic segmentation network, resulting in an integrated domain adaptation architecture we call HYLDA. To train the system end-to-end, we adopt a diverse set of learning paradigms, including 1) self-supervision on a simple auxiliary reconstruction task, 2) semi-supervised training using a few available labeled target domain frames, and 3) unsupervised training on the fake translated images generated by the imageto-image translation stage, together with the labeled frames from the source domain. In the latter case, the semantic segmentation network participates in the updating of the imageto-image translation engine. We demonstrate experimentally that HYLDA effectively addresses the challenging problem of improving generalization on validation data from the target domain when only a few target labeled frames are available for training. We perform an extensive evaluation where we compare HYLDA against strong baseline methods using two publicly available LiDAR semantic segmentation datasets.

show abstract

Domain Adaptation for Vehicle Detection from Bird's Eye View LiDAR Point Cloud Data

Cited by 49 publications

References 29 publications

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

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

Domain Adaptation on Point Clouds via Geometry-Aware Implicits

Domain Adaptation in LiDAR Semantic Segmentation via Alternating Skip Connections and Hybrid Learning

Contact Info

Product

Resources

About