Leveraging Temporal Data for Automatic Labelling of Static Vehicles

Walsh, Seán; Ku, Jason S.; Pon, Alex D.; Waslander, Steven L.

doi:10.1109/crv50864.2020.00026

Cited by 4 publications

(4 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[43] proposed an interesting technique that leverages differentiable 3D shape decoding and rendering to iteratively fit detailed 3D vehicle shapes to observed image and LiDAR evidence to generate automatic 3D labels, but only makes use of single frames, unlike our model. Finally, [33] restricts its domain to only consider detector-generated bounding boxes of static vehicles and generate a final label using the weighted sum of all associated detections. On the other hand, our work is able to seamlessly handle both dynamic and static objects through its flexible motion modeling.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Auto4D: Learning to Label 4D Objects from Sequential Point Clouds

Yang¹,

Liang²,

Zeng³

et al. 2021

Preprint

View full text Add to dashboard Cite

In the past few years we have seen great advances in 3D object detection thanks to deep learning methods. However, they typically rely on large amounts of high-quality labels to achieve good performance, which often require timeconsuming and expensive work by human annotators. To address this we propose an automatic annotation pipeline that generates accurate object trajectories in 3D (i.e., 4D labels) from LiDAR point clouds. Different from previous works that consider single frames at a time, our approach directly operates on sequential point clouds to combine richer object observations. The key idea is to decompose the 4D label into two parts: the 3D size of the object, and its motion path describing the evolution of the object's pose through time. More specifically, given a noisy but easy-toget object track as initialization, our model first estimates the object size from temporally aggregated observations, and then refines its motion path by considering both framewise observations as well as temporal motion cues. We validate the proposed method on a large-scale driving dataset and show that our approach achieves significant improvements over the baselines. We also showcase the benefits of our approach under the annotator-in-the-loop setting.

show abstract

Section: Related Workmentioning

confidence: 99%

“…For example, [26] pioneered the inclusion of handcrafted 3D motion priors to refine detections. [33] attempts to generate automatic 4D annotations by focusing on only static vehicles. However, because of their limitations in motion modeling capability or scenarios targeted, their application in automatic label generation is limited.…”

Section: Introductionmentioning

confidence: 99%

Auto4D: Learning to Label 4D Objects from Sequential Point Clouds

Yang¹,

Liang²,

Zeng³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…This mostly causes serious performance drops [50], which in turn leads to unreliable recognition systems. This can be mitigated by either manual or semi-supervised annotation [32,48] of representative data, each time the sensor setup or area of operation changes. However, this is infeasible for most real-world scenarios given the expensive labelling effort.…”

Section: Introductionmentioning

confidence: 99%

FAST3D: Flow-Aware Self-Training for 3D Object Detectors

Christian¹,

Opitz²,

Possegger³

et al. 2022

Preprint

View full text Add to dashboard Cite

In the field of autonomous driving, self-training is widely applied to mitigate distribution shifts in LiDAR-based 3D object detectors. This eliminates the need for expensive, high-quality labels whenever the environment changes (\eg, geographic location, sensor setup, weather condition). State-of-the-art self-training approaches, however, mostly ignore the temporal nature of autonomous driving data. % which contains additional information.To address this issue, we propose a flow-aware self-training method which enables unsupervised domain adaptation for 3D object detectors on continuous LiDAR point clouds. In order to get reliable pseudo-labels, we leverage scene flow to propagate detections through time. In particular, we introduce a flow-based multi-target tracker, that exploits flow consistency to filter and refine resulting tracks. The emerged precise pseudo-labels then serve as a basis for model re-training. Starting with a pre-trained KITTI model, we conduct experiments on the challenging Waymo Open Dataset to demonstrate the effectiveness of our approach. Without any prior target domain knowledge, our results show a significant improvement over the state-of-the-art.

show abstract

“…This mostly causes serious performance drops [50], which in turn leads to unreliable recognition systems. This can be mitigated by either manual or semi-supervised annotation [30,48] of representative data, each time the sensor setup or area of operation changes. However, this is infeasible for most real-world scenarios given the expensive labelling effort.…”

Section: Introductionmentioning

confidence: 99%

FAST3D: Flow-Aware Self-Training for 3D Object Detectors

Christian¹,

Opitz²,

Possegger³

et al. 2021

Preprint

View full text Add to dashboard Cite

In the field of autonomous driving, self-training is widely applied to mitigate distribution shifts in LiDAR-based 3D object detectors. This eliminates the need for expensive, high-quality labels whenever the environment changes (e.g., geographic location, sensor setup, weather condition). State-of-the-art self-training approaches, however, mostly ignore the temporal nature of autonomous driving data.To address this issue, we propose a flow-aware self-training method which enables unsupervised domain adaptation for 3D object detectors on continuous LiDAR point clouds. In order to get reliable pseudo-labels, we leverage scene flow to propagate detections through time. In particular, we introduce a flow-based multi-target tracker, that exploits flow consistency to filter and refine resulting tracks. The emerged precise pseudolabels then serve as a basis for model re-training. Starting with a pre-trained KITTI model, we conduct experiments on the challenging Waymo Open Dataset to demonstrate the effectiveness of our approach. Without any prior target domain knowledge, our results show a significant improvement over the state-of-the-art.

show abstract

Leveraging Temporal Data for Automatic Labelling of Static Vehicles

Cited by 4 publications

References 19 publications

Auto4D: Learning to Label 4D Objects from Sequential Point Clouds

Auto4D: Learning to Label 4D Objects from Sequential Point Clouds

FAST3D: Flow-Aware Self-Training for 3D Object Detectors

FAST3D: Flow-Aware Self-Training for 3D Object Detectors

Contact Info

Product

Resources

About