EvCenterNet: Uncertainty Estimation for Object Detection Using Evidential Learning

Uncertainty estimation is crucial in safety-critical settings such as automated driving as it provides valuable information for several downstream tasks including high-level decision making and path planning. In this work, we propose EvCenterNet, a novel uncertainty-aware 2D object detection framework using evidential learning to directly estimate both classification and regression uncertainties. To employ evidential learning for object detection, we devise a combination of evidential and focal loss functions for the sparse heatmap inputs. We introduce class-balanced weighting for regression and heatmap prediction to tackle the class imbalance encountered by evidential learning. Moreover, we propose a learning scheme to actively utilize the predicted heatmap uncertainties to improve the detection performance by focusing on the most uncertain points. We train our model on the KITTI dataset and evaluate it on challenging out-of-distribution datasets including BDD100K and nuImages. Our experiments demonstrate that our approach improves the precision and minimizes the execution time loss in relation to the base model.

Authors

• Monish R. Nallapareddy
• Kshitij Sirohi
• Paulo Lilles Jorge Drews
• Wolfram Burgard
• Chih-Hong Cheng
• Abhinav Valada

Keywords

• Heating systems
• Uncertainty
• Three-dimensional displays
• Decision making
• Estimation
• Focusing
• Object detection
• Uncertainty Estimation
• Evidential
• Detection Performance
• Path Planning
• Class Imbalance
• Focal Loss
• KITTI Dataset
• Convolutional Layers
• Binary Classification
• Pedestrian
• Bounding Box
• Additional Loss
• Central Objective
• Element-wise Multiplication
• 3D Convolution
• Dirichlet Distribution
• Epistemic Uncertainty
• Probable Point
• Frames Per Second
• 3D Convolutional Layers
• Detection Head
• Pixel Index
• Sparse Input
• Additional Hyperparameter
• Classification Uncertainty
• Object Height
• Digamma