Towards Better Performance and More Explainable Uncertainty for 3D Object Detection of Autonomous Vehicles

Pan, Hujie; Wang, Zining; Zhan, Wei; Tomizuka, Masayoshi

doi:10.1109/itsc45102.2020.9294177

Cited by 22 publications

(12 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…mean and variance in a Gaussian distribution). POD have been applied to RGB cameras [12], [36], [38], [39], LiDARs [11], [31], [37], [40], and Radars [41]. They have been shown to improve the detection robustness in the open-set conditions [12], reduce the labeling efforts in training [42], and enhance the detection accuracy [34], [36]- [39].…”

Section: Probabilistic Object Detection Networkmentioning

confidence: 99%

“…In this section, we use JIoU to quantitatively analyze how modelling uncertainties in BBoxes affects detection performance. To this end, we employ two state-of-the-art probabilistic object detectors based on LiDAR point clouds, namely, ProbPIXOR [35] and ProbPointRCNN [40]. The ProbPIXOR network models uncertainties from the PIXOR [48]…”

Section: B Evaluating Probabilistic Object Detectorsmentioning

confidence: 99%

“…ProbPIXOR ProbPointRCNN Fig. 13: A study of JIoU on two probabilistic object detectors ProbPIXOR [35] and ProbPointRCNN [40]. The gain of recall rates are measured by thresholding the JIoU scores from the detections with their predictive uncertainty estimates, compared to detections without uncertainty estimates.…”

Section: Recall Gain [%]mentioning

confidence: 99%

See 2 more Smart Citations

Labels Are Not Perfect: Inferring Spatial Uncertainty in Object Detection

Feng

Wang

Zhou

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

environmental noises in LiDAR perception and the label quality. Furthermore, we propose Jaccard IoU (JIoU) as a new evaluation metric that extends IoU by incorporating label uncertainty. We conduct an in-depth comparison among several LiDAR-based object detectors using the JIoU metric. Finally, we incorporate the proposed label uncertainty in a loss function to train a probabilistic object detector and to improve its detection accuracy. We verify our proposed methods on two public datasets (KITTI, Waymo), as well as on simulation data. Code is released at https://bit.ly/2W534yo.

show abstract

Section: Probabilistic Object Detection Networkmentioning

confidence: 99%

Section: B Evaluating Probabilistic Object Detectorsmentioning

confidence: 99%

See 1 more Smart Citation

Labels Are Not Perfect: Inferring Spatial Uncertainty in Object Detection

Feng

Wang

Zhou

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Zhu et al [3] offered a two-stage multi-modal fusion network for 3D object detection, the binocular images and raw point clouds can complement each other to improve the location accuracy. In [4], the authors proposed a new form of the loss function for the purpose of improving the performance of LiDAR-based 3D object detection and obtaining more explainable and convincing uncertainty for the prediction. However, the above detection strategies depend on highprecision LiDAR point cloud which cost is pretty high.…”

Section: Introductionmentioning

confidence: 99%

Monocular 3D Pedestrian Localization Fusing with Bird's Eye View

Chen

Zhang

Yuan

et al. 2021

2021 IEEE International Symposium on Circuits and Systems (ISCAS)

View full text Add to dashboard Cite

In recent years, 3D target detection and location methods in the field of autonomous driving have attracted increasing attention, but monocular 3D pedestrian localization research is still facing challenges. In this paper, a monocular pedestrian localization framework and a fine-grained location optimization method which is based on a bird's eye view are proposed. The monocular pedestrian localization framework is divided into three parts, which are coarse-grained localization, depth information reconstruction and fine-grained location optimization. In the stage of coarse-grained location, the human skeleton information is obtained from the original image by using the human skeleton point detection method, and then the pedestrian position is predicted through the method of a light-weight feed-forward neural network. In the stage of depth information reconstruction, the original image is used to reconstruct the corresponding bird's eye view with depth information through a parallel network. Finally, a fine-grained positioning optimization method makes it possible to get a more precise location with the help of the last two stages. The experimental results on the KITTI dataset show that our method has achieved better performance than the state-of-the-art methods.

show abstract

“…Previous studies have indicated that uncertain cases are strongly correlated with errors and identifying them increased model performance e.g. 17 There are some recent studies combining uncertainty and explainability information for 3D object detection 18 and clinical time serizes prediction. 19,20 Here, we use a random dropout methodology to approximate a Bayesian neural network to obtain a predictive posterior distribution and use this to quantify uncertainty in diagnosis.…”

Section: Introductionmentioning

confidence: 99%

Uncertainty aware and explainable diagnosis of retinal disease

Singh¹,

Sengupta²,

Rasheed³

et al. 2021

Preprint

View full text Add to dashboard Cite

Deep learning methods for ophthalmic diagnosis have shown considerable success in tasks like segmentation and classification. However, their widespread application is limited due to the models being opaque and vulnerable to making a wrong decision in complicated cases. Explainability methods show the features that a system used to make prediction while uncertainty awareness is the ability of a system to highlight when it is not sure about the decision. This is one of the first studies using uncertainty and explanations for informed clinical decision making. We perform uncertainty analysis of a deep learning model for diagnosis of four retinal diseases -age-related macular degeneration (AMD), central serous retinopathy (CSR), diabetic retinopathy (DR), and macular hole (MH) using images from a publicly available (OCTID) dataset. Monte Carlo (MC) dropout is used at the test time to generate a distribution of parameters and the predictions approximate the predictive posterior of a Bayesian model. A threshold is computed using the distribution and uncertain cases can be referred to the ophthalmologist thus avoiding an erroneous diagnosis. The features learned by the model are visualized using a proven attribution method from a previous study. The effects of uncertainty on model performance and the relationship between uncertainty and explainability are discussed in terms of clinical significance. The uncertainty information along with the heatmaps make the system more trustworthy for use in clinical settings.

show abstract

Towards Better Performance and More Explainable Uncertainty for 3D Object Detection of Autonomous Vehicles

Cited by 22 publications

References 20 publications

Labels Are Not Perfect: Inferring Spatial Uncertainty in Object Detection

Labels Are Not Perfect: Inferring Spatial Uncertainty in Object Detection

Monocular 3D Pedestrian Localization Fusing with Bird's Eye View

Uncertainty aware and explainable diagnosis of retinal disease

Contact Info

Product

Resources

About