RefinedMPL: Refined Monocular PseudoLiDAR for 3D Object Detection in Autonomous Driving

Vianney, Jean Marie Uwabeza; Aich, Shubhra; Liu, Bingbing

doi:10.48550/arxiv.1911.09712

Cited by 23 publications

(24 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although this problem is illposed, by obtaining a good prior knowledge of the surrounding scenes and objects it is possible to obtain reasonable results. Such methods were previously proposed by [1,6,10,11,20,26,27] to name a few. Here, we propose to combine elements from both monocular and stereo-reconstruction to derive our depth refinement scheme.…”

Section: Monocular Depth Estimationmentioning

confidence: 99%

Depth Refinement for Improved Stereo Reconstruction

Bracha¹,

Rotstein²,

David³

et al. 2021

Preprint

View full text Add to dashboard Cite

Depth estimation is a cornerstone of a vast number of applications requiring 3D assessment of the environment, such as robotics, augmented reality, and autonomous driving to name a few. One prominent technique for depth estimation is stereo matching which has several advantages: it is considered more accessible than other depth-sensing technologies, can produce dense depth estimates in realtime, and has benefited greatly from the advances of deep learning in recent years. However, current techniques for depth estimation from stereoscopic images still suffer from a built-in drawback. To reconstruct depth, a stereo matching algorithm first estimates the disparity map between the left and right images before applying a geometric triangulation. A simple analysis reveals that the depth error is quadratically proportional to the object's distance. Therefore, constant disparity errors are translated to large depth errors for objects far from the camera. To mitigate this quadratic relation, we propose a simple but effective method that uses a refinement network for depth estimation. We show analytical and empirical results suggesting that the proposed learning procedure reduces this quadratic relation. We evaluate the proposed refinement procedure on well-known benchmarks and datasets, like Sceneflow and KITTI datasets, and demonstrate significant improvements in the depth accuracy metric.

show abstract

Section: Monocular Depth Estimationmentioning

confidence: 99%

Depth Refinement for Improved Stereo Reconstruction

Bracha¹,

Rotstein²,

David³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Reconstructing spatial information is the core issue for monocular 3D object detection. Some methods [16,17,[25][26][27]29] rely on the existing monocular depth estimation algorithms. Pseudo-LIDAR [26] transforms the estimated depth image to artificial dense point clouds so as to employ LIDAR-based 3D object detectors [7,19].…”

Section: Related Workmentioning

confidence: 99%

“…anchor-based and keypoint-based methods. Pseudo-LiDAR related methods [16,17,[25][26][27]29]acquire depth image using the existing monocular depth estimation algorithm and transform the acquired depth image to pseudo-LiDAR point cloud which can be fed to LIDAR-based 3D object detectors. However, these methods usually suffer from poor efficiency due to the additional depth estimation module.…”

Section: Introductionmentioning

confidence: 99%

Lite-FPN for Keypoint-based Monocular 3D Object Detection

Yang¹,

Zhang²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

3D object detection with a single image is an essential and challenging task for autonomous driving. Recently, keypoint-based monocular 3D object detection has made tremendous progress and achieved great speed-accuracy trade-off. However, there still exists a huge gap with LIDAR-based methods in terms of accuracy. To improve their performance without sacrificing efficiency, we propose a sort of lightweight feature pyramid network called Lite-FPN to achieve multi-scale feature fusion in an effective and efficient way, which can boost the multi-scale detection capability of keypoint-based detectors. Besides, the misalignment between the classification score and the localization precision is further relieved by introducing a novel regression loss named attention loss. With the proposed loss, predictions with high confidence but poor localization are treated with more attention during the training phase. Comparative experiments based on several state-ofthe-art keypoint-based detectors on the KITTI dataset show that our proposed method achieves significantly higher accuracy and frame rate at the same time. The code and pretrained models will be available at https://github. com/yanglei18/Lite-FPN .

show abstract

“…For instance, [5,2] utilize CAD models to assist in estimating the depth of the vehicle. Similarly, a pre-trained depth estimation model is adopted to estimate the depth information of the scene in [37,1,40]. However, such methods directly or indirectly used 3D depth ground-truth data in monocular 3D object detection.…”

Section: Introductionmentioning

confidence: 99%

M3DSSD: Monocular 3D Single Stage Object Detector

Luo

Dai

Shao

et al. 2021

2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

In this paper, we propose a Monocular 3D Single Stage object Detector (M3DSSD) with feature alignment and asymmetric non-local attention. Current anchor-based monocular 3D object detection methods suffer from feature mismatching. To overcome this, we propose a twostep feature alignment approach. In the first step, the shape alignment is performed to enable the receptive field of the feature map to focus on the pre-defined anchors with high confidence scores. In the second step, the center alignment is used to align the features at 2D/3D centers. Further, it is often difficult to learn global information and capture long-range relationships, which are important for the depth prediction of objects. Therefore, we propose a novel asymmetric non-local attention block with multiscale sampling to extract depth-wise features. The proposed M3DSSD achieves significantly better performance than the monocular 3D object detection methods on the KITTI dataset, in both 3D object detection and bird's eye view tasks. The code is released at https://github. com/mumianyuxin/M3DSSD.

show abstract

RefinedMPL: Refined Monocular PseudoLiDAR for 3D Object Detection in Autonomous Driving

Cited by 23 publications

References 45 publications

Depth Refinement for Improved Stereo Reconstruction

Depth Refinement for Improved Stereo Reconstruction

Lite-FPN for Keypoint-based Monocular 3D Object Detection

M3DSSD: Monocular 3D Single Stage Object Detector

Contact Info

Product

Resources

About