MonoPair: Monocular 3D Object Detection Using Pairwise Spatial Relationships

Chen, Yongjian; Tai, Lei; Sun, Kai; Li, Mingyang

doi:10.1109/cvpr42600.2020.01211

Cited by 256 publications

(167 citation statements)

References 30 publications

Supporting

Mentioning

167

Contrasting

Order By: Relevance

“…Table 24 presents that, for the pedestrian category, the results of [55,66] outperform other methods on all three levels for AP bev and AP 3D (IoU @ 0.7) and for AP bev and AP 3D (IoU @ 0.5) respectively. For cyclist category the results of [69] and third method of [57] have higher performance on all three levels when compared using AP bev and AP 3D (IoU @ 0.5) and AP bev and AP 3D (IoU @ 0.7).…”

Section: Discussionmentioning

confidence: 99%

“…Even though existing 3D detectors have achieved good accuracy, most of them do not consider the information related to occluded objects, which are partially visible. To this end, Chen et al [55] improved 3D object detection by establishing a relationship of paired samples, which allows modeling spatial constraints for occluded objects. Its 3D detector introduced an uncertainty-aware prediction module for computing object location and object-to-object distances.…”

Section: (I) Monocular-based 3dormentioning

confidence: 99%

“…For performance evaluation based on the KITTI dataset, Mono Pair [55] uses 40-point interpolated average precision metric AP 40 , which is evaluated at both the bird-eye view AP bev and the 3D bounding box AP 3d . It reports AP with intersection over union (IoU) using 0.7 as thresholds for cars, pedestrians and cyclists detection.…”

Section: Performance Evaluationmentioning

confidence: 99%

“…It reports AP with intersection over union (IoU) using 0.7 as thresholds for cars, pedestrians and cyclists detection. Tables 21 and 22 shows the performance of one-stage anchor-free detector of [55] on the KITTI validation and test sets for the car category, while performance for pedestrians and cyclists on the KITTI test is shown in Tables 23 and 24, respectively. It can also perform inference in real-time as 57 ms per image, which is higher than [106].…”

Section: Performance Evaluationmentioning

confidence: 99%

See 3 more Smart Citations

3D Recognition Based on Sensor Modalities for Robotic Systems: A Survey

Manzoor

Joo

Kim

et al. 2021

Sensors

View full text Add to dashboard Cite

3D visual recognition is a prerequisite for most autonomous robotic systems operating in the real world. It empowers robots to perform a variety of tasks, such as tracking, understanding the environment, and human–robot interaction. Autonomous robots equipped with 3D recognition capability can better perform their social roles through supportive task assistance in professional jobs and effective domestic services. For active assistance, social robots must recognize their surroundings, including objects and places to perform the task more efficiently. This article first highlights the value-centric role of social robots in society by presenting recently developed robots and describes their main features. Instigated by the recognition capability of social robots, we present the analysis of data representation methods based on sensor modalities for 3D object and place recognition using deep learning models. In this direction, we delineate the research gaps that need to be addressed, summarize 3D recognition datasets, and present performance comparisons. Finally, a discussion of future research directions concludes the article. This survey is intended to show how recent developments in 3D visual recognition based on sensor modalities using deep-learning-based approaches can lay the groundwork to inspire further research and serves as a guide to those who are interested in vision-based robotics applications.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: (I) Monocular-based 3dormentioning

confidence: 99%

Section: Performance Evaluationmentioning

confidence: 99%

Section: Performance Evaluationmentioning

confidence: 99%

See 2 more Smart Citations

3D Recognition Based on Sensor Modalities for Robotic Systems: A Survey

Manzoor

Joo

Kim

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Unfortunately, the solution of the 3D position depends on the results of previous processes, which will cause the errors to be further amplified and affect the final detection results. Notably, Chen et al [10] propose to utilize the pairwise relationship of the objects as a constraint to characterize the errors that exist in the detection pipeline. This method greatly improves the detection capability.…”

Section: Introductionmentioning

confidence: 99%

Monocular 3D Object Detection Based on Uncertainty Prediction of Keypoints

Zhao

Liu

2021

Machines

View full text Add to dashboard Cite

Three-dimensional (3D) object detection is an important task in the field of machine vision, in which the detection of 3D objects using monocular vision is even more challenging. We observe that most of the existing monocular methods focus on the design of the feature extraction framework or embedded geometric constraints, but ignore the possible errors in the intermediate process of the detection pipeline. These errors may be further amplified in the subsequent processes. After exploring the existing detection framework of keypoints, we find that the accuracy of keypoints prediction will seriously affect the solution of 3D object position. Therefore, we propose a novel keypoints uncertainty prediction network (KUP-Net) for monocular 3D object detection. In this work, we design an uncertainty prediction module to characterize the uncertainty that exists in keypoint prediction. Then, the uncertainty is used for joint optimization with object position. In addition, we adopt position-encoding to assist the uncertainty prediction, and use a timing coefficient to optimize the learning process. The experiments on our detector are conducted on the KITTI benchmark. For the two levels of easy and moderate, we achieve accuracy of 17.26 and 11.78 in AP3D, and achieve accuracy of 23.59 and 16.63 in APBEV, which are higher than the latest method KM3D.

show abstract