Improving Video Instance Segmentation by Light-weight Temporal Uncertainty Estimates

Maag, Kira; Rottmann, Matthias; Varghese, Serin; Hüger, Fabian; Schlicht, Peter; Gottschalk, Hanno

doi:10.1109/ijcnn52387.2021.9534407

Cited by 7 publications

(14 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To reduce false negative instance predictions, spatial and motion information shared by all instances is captured. The method presented in [6] compares the ordinary score thresholding during inference instance segmentation instance tracking false negative detection meta classification depth estimation Fig. 2.…”

Section: Related Workmentioning

confidence: 99%

“…Neural networks as statistical models produce probabilistic predictions prone to error, for this reason, it is necessary to understand and minimize these errors. In safety critical applications like automated driving [1] and medical diagnosis [2], the reliability of neural networks in terms of uncertainty quantification [3] and prediction quality estimation [4]- [6] is of highest interest. Instance segmentation networks (for example Mask R-CNN [7] and YOLACT [8]) provide for each object a confidence value, also called score.…”

Section: Introductionmentioning

confidence: 99%

“…By this detection method, the number of instances can be greatly increased and thus, we deploy meta classification to improve also the precision rate. Meta classification for instance segmentation is presented in [6] and was previously introduced for semantic segmentation in [11]. Meta classification addresses the task of predicting if a predicted instance intersects with ground truth or not.…”

Section: Introductionmentioning

confidence: 99%

“…Based on the resulting heatmap, we construct further metrics aggregated per instance. We complete our set of metrics with a few measures presented in [6] that are based on expected position, changes in the shape and survival time analysis of instances in an image sequence.…”

Section: Introductionmentioning

confidence: 99%

“…• We introduce a light-weight detection algorithm for predicted instances obtained by a neural network and tracked by the algorithm of [6]. Furthermore, the time-dynamic metrics which serve as input for the meta classification model are defined.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

False Negative Reduction in Video Instance Segmentation using Uncertainty Estimates

Maag¹

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Instance segmentation of images is an important tool for automated scene understanding. Neural networks are usually trained to optimize their overall performance in terms of accuracy. Meanwhile, in applications such as automated driving, an overlooked pedestrian seems more harmful than a falsely detected one. In this work, we present a false negative detection method for image sequences based on inconsistencies in time series of tracked instances given the availability of image sequences in online applications. As the number of instances can be greatly increased by this algorithm, we apply a false positive pruning using uncertainty estimates aggregated over instances. To this end, instance-wise metrics are constructed which characterize uncertainty and geometry of a given instance or are predicated on depth estimation. The proposed method serves as a post-processing step applicable to any neural network that can also be trained on single frames only. In our tests, we obtain an improved trade-off between false negative and false positive instances by our fused detection approach in comparison to the use of an ordinary score value provided by the instance segmentation network during inference.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

False Negative Reduction in Video Instance Segmentation using Uncertainty Estimates

Maag¹

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Uncertainty Quantification for Object Detection: Output- and Gradient-Based Approaches

Riedlinger

Schubert

Kahl

et al. 2022

Deep Neural Networks and Data for Automated Driving

View full text Add to dashboard Cite

Safety-critical applications of deep neural networks require reliable confidence estimation methods with high predictive power. However, evaluating and comparing different methods for uncertainty quantification is oftentimes highly context-dependent. In this chapter, we introduce flexible evaluation protocols which are applicable to a wide range of tasks with an emphasis on object detection. In this light, we investigate uncertainty metrics based on the network output, as well as metrics based on a learning gradient, both of which significantly outperform the confidence score of the network. While output-based uncertainty is produced by post-processing steps and is computationally efficient, gradient-based uncertainty, in principle, allows for localization of uncertainty within the network architecture. We show that both sources of uncertainty are mutually non-redundant and can be combined beneficially. Furthermore, we show direct applications of uncertainty quantification by improving detection accuracy.

show abstract

Two Video Data Sets for Tracking and Retrieval of Out of Distribution Objects

Maag

Chan

Svenja

et al. 2023

Computer Vision – ACCV 2022

View full text Add to dashboard Cite

Improving Video Instance Segmentation by Light-weight Temporal Uncertainty Estimates

Cited by 7 publications

References 28 publications

False Negative Reduction in Video Instance Segmentation using Uncertainty Estimates

False Negative Reduction in Video Instance Segmentation using Uncertainty Estimates

Uncertainty Quantification for Object Detection: Output- and Gradient-Based Approaches

Two Video Data Sets for Tracking and Retrieval of Out of Distribution Objects

Contact Info

Product

Resources

About