Gradient-Based Quantification of Epistemic Uncertainty for Deep Object Detectors

Abstract

Reliable epistemic uncertainty estimation is an essential component for backend applications of deep object detectors in safety-critical environments. Modern network architectures tend to give poorly calibrated confidences with limited predictive power. Here, we introduce novel gradient-based uncertainty metrics and investigate them for different object detection architectures. Experiments on the MS COCO, PASCAL VOC and the KITTI dataset show significant improvements in true positive / false positive discrimination and prediction of intersection over union as compared to network confidence. We also find improvement over Monte-Carlo dropout uncertainty metrics and further significant boosts by aggregating different sources of uncertainty metrics. The resulting uncertainty models generate well-calibrated confidences in all instances. Furthermore, we implement our uncertainty quantification models into object detection pipelines as a means to discern true against false predictions, replacing the ordinary score-threshold-based decision rule. In our experiments, we achieve a significant boost in detection performance in terms of mean average precision. With respect to computational complexity, we find that computing gradient uncertainty metrics results in floating point operation counts similar to those of Monte-Carlo dropout.

Uncertainty Quantification and Resource-Demanding Computer Vision Applications of Deep Learning

“…Our proposed active learning strategy builds upon previous work in uncertainty quantification for deep object detection 20,21 . The MetaDetect 20 and gradient uncertainty 21 approaches have shown state-of-the-art uncertainty quantification in terms of false positive detection of individual predicted instances as well, as localisation uncertainty estimation.…”

“…Our proposed active learning strategy builds upon previous work in uncertainty quantification for deep object detection 20,21 . The MetaDetect 20 and gradient uncertainty 21 approaches have shown state-of-the-art uncertainty quantification in terms of false positive detection of individual predicted instances as well, as localisation uncertainty estimation. Moreover, the proposed methods do not affect the training process of the neural network, therefore, fitting as a post-processing module fitted on a validation dataset V on top of any object detector making the implementation into an active learning cycle flexible.…”

“…In the past years, UQ for image segmentation and object detection has been initiated in a series of works [16][17][18][19][20][21] focusing on false positive instances and false negative instances 22,23 , respectively, we also refer to a survey 17 . While many works focus on street scene recognition for autonomous driving, 16,24 cover medical image segmentation applications as well.…”