Matthias Rottmann

Marius Schubert, Tobias Riedlinger, Karsten Kahl et al.

Labeling datasets for supervised object detection is a dull and time-consuming task. Errors can be easily introduced during annotation and overlooked during review, yielding inaccurate benchmarks and performance degradation of deep neural networks trained on noisy labels. In this work, we for the first time introduce a benchmark for label error detection methods on object detection datasets as well as a label error detection method and a number of baselines. We simulate four different types of randomly introduced label errors on train and test sets of well-labeled object detection datasets. For our label error detection method we assume a two-stage object detector to be given and consider the sum of both stages' classification and regression losses. The losses are computed with respect to the predictions and the noisy labels including simulated label errors, aiming at detecting the latter. We compare our method to three baselines: a naive one without deep learning, the object detector's score and the entropy of the classification softmax distribution. We outperform all baselines and demonstrate that among the considered methods, ours is the only one that detects label errors of all four types efficiently. Furthermore, we detect real label errors a) on commonly used test datasets in object detection and b) on a proprietary dataset. In both cases we achieve low false positives rates, i.e., we detect label errors with a precision for a) of up to 71.5% and for b) with 97%.

Matthias Rottmann, Marco Reese

In this work, we for the first time present a method for detecting label errors in image datasets with semantic segmentation, i.e., pixel-wise class labels. Annotation acquisition for semantic segmentation datasets is time-consuming and requires plenty of human labor. In particular, review processes are time consuming and label errors can easily be overlooked by humans. The consequences are biased benchmarks and in extreme cases also performance degradation of deep neural networks (DNNs) trained on such datasets. DNNs for semantic segmentation yield pixel-wise predictions, which makes detection of label errors via uncertainty quantification a complex task. Uncertainty is particularly pronounced at the transitions between connected components of the prediction. By lifting the consideration of uncertainty to the level of predicted components, we enable the usage of DNNs together with component-level uncertainty quantification for the detection of label errors. We present a principled approach to benchmarking the task of label error detection by dropping labels from the Cityscapes dataset as well from a dataset extracted from the CARLA driving simulator, where in the latter case we have the labels under control. Our experiments show that our approach is able to detect the vast majority of label errors while controlling the number of false label error detections. Furthermore, we apply our method to semantic segmentation datasets frequently used by the computer vision community and present a collection of label errors along with sample statistics.

Robin Chan, Radin Dardashti, Meike Osinski et al.

Deep neural networks (DNN) have made impressive progress in the interpretation of image data, so that it is conceivable and to some degree realistic to use them in safety critical applications like automated driving. From an ethical standpoint, the AI algorithm should take into account the vulnerability of objects or subjects on the street that ranges from "not at all", e.g. the road itself, to "high vulnerability" of pedestrians. One way to take this into account is to define the cost of confusion of one semantic category with another and use cost-based decision rules for the interpretation of probabilities, which are the output of DNNs. However, it is an open problem how to define the cost structure, who should be in charge to do that, and thereby define what AI-algorithms will actually "see". As one possible answer, we follow a participatory approach and set up an online survey to ask the public to define the cost structure. We present the survey design and the data acquired along with an evaluation that also distinguishes between perspective (car passenger vs. external traffic participant) and gender. Using simulation based $F$-tests, we find highly significant differences between the groups. These differences have consequences on the reliable detection of pedestrians in a safety critical distance to the self-driving car. We discuss the ethical problems that are related to this approach and also discuss the problems emerging from human-machine interaction through the survey from a psychological point of view. Finally, we include comments from industry leaders in the field of AI safety on the applicability of survey based elements in the design of AI functionalities in automated driving.

Julian Burghoff, Matthias Rottmann, Jill von Conta et al.

In this work, we develop a neural architecture search algorithm, termed Resbuilder, that develops ResNet architectures from scratch that achieve high accuracy at moderate computational cost. It can also be used to modify existing architectures and has the capability to remove and insert ResNet blocks, in this way searching for suitable architectures in the space of ResNet architectures. In our experiments on different image classification datasets, Resbuilder achieves close to state-of-the-art performance while saving computational cost compared to off-the-shelf ResNets. Noteworthy, we once tune the parameters on CIFAR10 which yields a suitable default choice for all other datasets. We demonstrate that this property generalizes even to industrial applications by applying our method with default parameters on a proprietary fraud detection dataset.

Karsten Kahl, Matthias Rottmann

The bootstrap algebraic multigrid framework allows for the adaptive construction of algebraic multigrid methods in situations where geometric multigrid methods are not known or not available at all. While there has been some work on adaptive coarsening in this framework in terms of algebraic distances, coarsening is the part of the adaptive bootstrap setup that is least developed. In this paper we try to close this gap by introducing an adaptive coarsening scheme that views interpolation as a local regression problem. In fact the bootstrap algebraic multigrid setup can be understood as a machine learning ansatz that learns the nature of smooth error by local regression. In order to turn this idea into a practical method we modify least squares interpolation to both avoid overfitting of the data and to recover a sparse response that can be used to extract information about the coupling strength amongst variables like in classical algebraic multigrid. In order to improve the so-found coarse grid we propose a post-processing to ensure stability of the resulting least squares interpolation operator. We conclude with numerical experiments that show the viability of the chosen approach.

Annika Mütze, Matthias Rottmann, Hanno Gottschalk

Domain adaptation is of huge interest as labeling is an expensive and error-prone task, especially when labels are needed on pixel-level like in semantic segmentation. Therefore, one would like to be able to train neural networks on synthetic domains, where data is abundant and labels are precise. However, these models often perform poorly on out-of-domain images. To mitigate the shift in the input, image-to-image approaches can be used. Nevertheless, standard image-to-image approaches that bridge the domain of deployment with the synthetic training domain do not focus on the downstream task but only on the visual inspection level. We therefore propose a "task aware" version of a GAN in an image-to-image domain adaptation approach. With the help of a small amount of labeled ground truth data, we guide the image-to-image translation to a more suitable input image for a semantic segmentation network trained on synthetic data (synthetic-domain expert). The main contributions of this work are 1) a modular semi-supervised domain adaptation method for semantic segmentation by training a downstream task aware CycleGAN while refraining from adapting the synthetic semantic segmentation expert 2) the demonstration that the method is applicable to complex domain adaptation tasks and 3) a less biased domain gap analysis by using from scratch networks. We evaluate our method on a classification task as well as on semantic segmentation. Our experiments demonstrate that our method outperforms CycleGAN - a standard image-to-image approach - by 7 percent points in accuracy in a classification task using only 70 (10%) ground truth images. For semantic segmentation we can show an improvement of about 4 to 7 percent points in mean Intersection over union on the Cityscapes evaluation dataset with only 14 ground truth images during training.

Marius Schubert, Tobias Riedlinger, Karsten Kahl et al.

Obtaining annotations for complex computer vision tasks such as object detection is an expensive and time-intense endeavor involving a large number of human workers or expert opinions. Reducing the amount of annotations required while maintaining algorithm performance is, therefore, desirable for machine learning practitioners and has been successfully achieved by active learning algorithms. However, it is not merely the amount of annotations which influences model performance but also the annotation quality. In practice, the oracles that are queried for new annotations frequently contain significant amounts of noise. Therefore, cleansing procedures are oftentimes necessary to review and correct given labels. This process is subject to the same budget as the initial annotation itself since it requires human workers or even domain experts. Here, we propose a composite active learning framework including a label review module for deep object detection. We show that utilizing part of the annotation budget to correct the noisy annotations partially in the active dataset leads to early improvements in model performance, especially when coupled with uncertainty-based query strategies. The precision of the label error proposals has a significant influence on the measured effect of the label review. In our experiments we achieve improvements of up to 4.5 mAP points of object detection performance by incorporating label reviews at equal annotation budget.

Tobias Riedlinger, Marius Schubert, Sarina Penquitt et al.

Object detection on Lidar point cloud data is a promising technology for autonomous driving and robotics which has seen a significant rise in performance and accuracy during recent years. Particularly uncertainty estimation is a crucial component for down-stream tasks and deep neural networks remain error-prone even for predictions with high confidence. Previously proposed methods for quantifying prediction uncertainty tend to alter the training scheme of the detector or rely on prediction sampling which results in vastly increased inference time. In order to address these two issues, we propose LidarMetaDetect (LMD), a light-weight post-processing scheme for prediction quality estimation. Our method can easily be added to any pre-trained Lidar object detector without altering anything about the base model and is purely based on post-processing, therefore, only leading to a negligible computational overhead. Our experiments show a significant increase of statistical reliability in separating true from false predictions. We propose and evaluate an additional application of our method leading to the detection of annotation errors. Explicit samples and a conservative count of annotation error proposals indicates the viability of our method for large-scale datasets like KITTI and nuScenes. On the widely-used nuScenes test dataset, 43 out of the top 100 proposals of our method indicate, in fact, erroneous annotations.

Tobias Riedlinger, Marius Schubert, Karsten Kahl et al.

Active learning as a paradigm in deep learning is especially important in applications involving intricate perception tasks such as object detection where labels are difficult and expensive to acquire. Development of active learning methods in such fields is highly computationally expensive and time consuming which obstructs the progression of research and leads to a lack of comparability between methods. In this work, we propose and investigate a sandbox setup for rapid development and transparent evaluation of active learning in deep object detection. Our experiments with commonly used configurations of datasets and detection architectures found in the literature show that results obtained in our sandbox environment are representative of results on standard configurations. The total compute time to obtain results and assess the learning behavior can thereby be reduced by factors of up to 14 when comparing with Pascal VOC and up to 32 when comparing with BDD100k. This allows for testing and evaluating data acquisition and labeling strategies in under half a day and contributes to the transparency and development speed in the field of active learning for object detection.

Kira Maag, Matthias Rottmann

State-of-the-art deep neural networks demonstrate outstanding performance in semantic segmentation. However, their performance is tied to the domain represented by the training data. Open world scenarios cause inaccurate predictions which is hazardous in safety relevant applications like automated driving. In this work, we enhance semantic segmentation predictions using monocular depth estimation to improve segmentation by reducing the occurrence of non-detected objects in presence of domain shift. To this end, we infer a depth heatmap via a modified segmentation network which generates foreground-background masks, operating in parallel to a given semantic segmentation network. Both segmentation masks are aggregated with a focus on foreground classes (here road users) to reduce false negatives. To also reduce the occurrence of false positives, we apply a pruning based on uncertainty estimates. Our approach is modular in a sense that it post-processes the output of any semantic segmentation network. In our experiments, we observe less non-detected objects of most important classes and an enhanced generalization to other domains compared to the basic semantic segmentation prediction.

Antonia van Betteray, Matthias Rottmann, Karsten Kahl

Current state-of-the-art deep neural networks for image classification are made up of 10 - 100 million learnable weights and are therefore inherently prone to overfitting. The complexity of the weight count can be seen as a function of the number of channels, the spatial extent of the input and the number of layers of the network. Due to the use of convolutional layers the scaling of weight complexity is usually linear with regards to the resolution dimensions, but remains quadratic with respect to the number of channels. Active research in recent years in terms of using multigrid inspired ideas in deep neural networks have shown that on one hand a significant number of weights can be saved by appropriate weight sharing and on the other that a hierarchical structure in the channel dimension can improve the weight complexity to linear. In this work, we combine these multigrid ideas to introduce a joint framework of multigrid inspired architectures, that exploit multigrid structures in all relevant dimensions to achieve linear weight complexity scaling and drastically reduced weight counts. Our experiments show that this structured reduction in weight count is able to reduce overfitting and thus shows improved performance over state-of-the-art ResNet architectures on typical image classification benchmarks at lower network complexity.

Edgar Heinert, Stephan Tilgner, Timo Palm et al.

When employing deep neural networks (DNNs) for semantic segmentation in safety-critical applications like automotive perception or medical imaging, it is important to estimate their performance at runtime, e.g. via uncertainty estimates or prediction quality estimates. Previous works mostly performed uncertainty estimation on pixel-level. In a line of research, a connected-component-wise (segment-wise) perspective was taken, approaching uncertainty estimation on an object-level by performing so-called meta classification and regression to estimate uncertainty and prediction quality, respectively. In those works, each predicted segment is considered individually to estimate its uncertainty or prediction quality. However, the neighboring segments may provide additional hints on whether a given predicted segment is of high quality, which we study in the present work. On the basis of uncertainty indicating metrics on segment-level, we use graph neural networks (GNNs) to model the relationship of a given segment's quality as a function of the given segment's metrics as well as those of its neighboring segments. We compare different GNN architectures and achieve a notable performance improvement.

Julian Burghoff, Robin Chan, Hanno Gottschalk et al.

Bringing deep neural networks (DNNs) into safety critical applications such as automated driving, medical imaging and finance, requires a thorough treatment of the model's uncertainties. Training deep neural networks is already resource demanding and so is also their uncertainty quantification. In this overview article, we survey methods that we developed to teach DNNs to be uncertain when they encounter new object classes. Additionally, we present training methods to learn from only a few labels with help of uncertainty quantification. Note that this is typically paid with a massive overhead in computation of an order of magnitude and more compared to ordinary network training. Finally, we survey our work on neural architecture search which is also an order of magnitude more resource demanding then ordinary network training.

Gesina Schwalbe, Mert Keser, Moritz Bayerkuhnlein et al.

Vision-language model (VLM) encoders such as CLIP enable strong retrieval and zero-shot classification in a shared image-text embedding space, yet the semantic organization of this space is rarely inspected. We present a post-hoc framework to explain, verify, and align the semantic hierarchies induced by a VLM over a given set of child classes. First, we extract a binary hierarchy by agglomerative clustering of class centroids and name internal nodes by dictionary-based matching to a concept bank. Second, we quantify plausibility by comparing the extracted tree against human ontologies using efficient tree- and edge-level consistency measures, and we evaluate utility via explainable hierarchical tree-traversal inference with uncertainty-aware early stopping (UAES). Third, we propose an ontology-guided post-hoc alignment method that learns a lightweight embedding-space transformation, using UMAP to generate target neighborhoods from a desired hierarchy. Across 13 pretrained VLMs and 4 image datasets, our method finds systematic modality differences: image encoders are more discriminative, while text encoders induce hierarchies that better match human taxonomies. Overall, the results reveal a persistent trade-off between zero-shot accuracy and ontological plausibility and suggest practical routes to improve semantic alignment in shared embedding spaces.

Krzysztof Lis, Matthias Rottmann, Annika Mütze et al.

In addition to impressive performance, vision transformers have demonstrated remarkable abilities to encode information they were not trained to extract. For example, this information can be used to perform segmentation or single-view depth estimation even though the networks were only trained for image recognition. We show that a similar phenomenon occurs when explicitly training transformers for semantic segmentation in a supervised manner for a set of categories: Once trained, they provide valuable information even about categories absent from the training set. This information can be used to segment objects from these never-seen-before classes in domains as varied as road obstacles, aircraft parked at a terminal, lunar rocks, and maritime hazards.

Mert Keser, Gesina Schwalbe, Niki Amini-Naieni et al.

Ontological commitment, i.e., used concepts, relations, and assumptions, are a corner stone of qualitative reasoning (QR) models. The state-of-the-art for processing raw inputs, though, are deep neural networks (DNNs), nowadays often based off from multimodal foundation models. These automatically learn rich representations of concepts and respective reasoning. Unfortunately, the learned qualitative knowledge is opaque, preventing easy inspection, validation, or adaptation against available QR models. So far, it is possible to associate pre-defined concepts with latent representations of DNNs, but extractable relations are mostly limited to semantic similarity. As a next step towards QR for validation and verification of DNNs: Concretely, we propose a method that extracts the learned superclass hierarchy from a multimodal DNN for a given set of leaf concepts. Under the hood we (1) obtain leaf concept embeddings using the DNN's textual input modality; (2) apply hierarchical clustering to them, using that DNNs encode semantic similarities via vector distances; and (3) label the such-obtained parent concepts using search in available ontologies from QR. An initial evaluation study shows that meaningful ontological class hierarchies can be extracted from state-of-the-art foundation models. Furthermore, we demonstrate how to validate and verify a DNN's learned representations against given ontologies. Lastly, we discuss potential future applications in the context of QR.

Chamuditha Jayanga Galappaththige, Thomas Gottwald, Peter Stehr et al.

Recent advances in 3D Gaussian Splatting have enabled impressive photorealistic novel view synthesis. However, to transition from a pure rendering engine to a reliable spatial map for autonomous agents and safety-critical applications, knowing where the representation is uncertain is as important as the rendering fidelity itself. We bridge this critical gap by introducing a lightweight, plug-and-play framework for pixel-wise, view-dependent predictive uncertainty estimation. Our post-hoc method formulates uncertainty as a Bayesian-regularized linear least-squares optimization over reconstruction residuals. This architecture-agnostic approach extracts a per-primitive uncertainty channel without modifying the underlying scene representation or degrading baseline visual fidelity. Crucially, we demonstrate that providing this actionable reliability signal successfully translates 3D Gaussian splatting into a trustworthy spatial map, further improving state-of-the-art performance across three critical downstream perception tasks: active view selection, pose-agnostic scene change detection, and pose-agnostic anomaly detection.

Sadia Ilyas, Ido Freeman, Matthias Rottmann

Out-of-distribution (OOD) object detection is a critical task focused on detecting objects that originate from a data distribution different from that of the training data. In this study, we investigate to what extent state-of-the-art open-vocabulary object detectors can detect unusual objects in street scenes, which are considered as OOD or rare scenarios with respect to common street scene datasets. Specifically, we evaluate their performance on the OoDIS Benchmark, which extends RoadAnomaly21 and RoadObstacle21 from SegmentMeIfYouCan, as well as LostAndFound, which was recently extended to object level annotations. The objective of our study is to uncover short-comings of contemporary object detectors in challenging real-world, and particularly in open-world scenarios. Our experiments reveal that open vocabulary models are promising for OOD object detection scenarios, however far from perfect. Substantial improvements are required before they can be reliably deployed in real-world applications. We benchmark four state-of-the-art open-vocabulary object detection models on three different datasets. Noteworthily, Grounding DINO achieves the best results on RoadObstacle21 and LostAndFound in our study with an AP of 48.3% and 25.4% respectively. YOLO-World excels on RoadAnomaly21 with an AP of 21.2%.

Sadia Ilyas, Annika Mütze, Klaus Friedrichs et al.

Out-of-distribution (OOD) object detection is an important yet underexplored task. A reliable object detector should be able to handle OOD objects by localizing and correctly classifying them as OOD. However, a critical issue arises when such atypical objects are completely missed by the object detector and incorrectly treated as background. Existing OOD detection approaches in object detection often rely on complex architectures or auxiliary branches and typically do not provide a framework that treats in-distribution (ID) and OOD in a unified way. In this work, we address these limitations by enabling a single detector to detect OOD objects, that are otherwise silently overlooked, alongside ID objects. We present \textbf{SynOE-OD}, a \textbf{Syn}thetic \textbf{O}utlier-\textbf{E}xposure-based \textbf{O}bject \textbf{D}etection framework, that leverages strong generative models, like Stable Diffusion, and Open-Vocabulary Object Detectors (OVODs) to generate semantically meaningful, object-level data that serve as outliers during training. The generated data is used for transfer-learning to establish strong ID task performance and supplement detection models with OOD object detection robustness. Our approach achieves state-of-the-art average precision on an established OOD object detection benchmark, where OVODs, such as GroundingDINO, show limited zero-shot performance in detecting OOD objects in street-scenes.

Lorenc Kapllani, Long Teng, Matthias Rottmann

Deep learning-based numerical schemes for solving high-dimensional backward stochastic differential equations (BSDEs) have recently raised plenty of scientific interest. While they enable numerical methods to approximate very high-dimensional BSDEs, their reliability has not been studied and is thus not understood. In this work, we study uncertainty quantification (UQ) for a class of deep learning-based BSDE schemes. More precisely, we review the sources of uncertainty involved in the schemes and numerically study the impact of different sources. Usually, the standard deviation (STD) of the approximate solutions obtained from multiple runs of the algorithm with different datasets is calculated to address the uncertainty. This approach is computationally quite expensive, especially for high-dimensional problems. Hence, we develop a UQ model that efficiently estimates the STD of the approximate solution using only a single run of the algorithm. The model also estimates the mean of the approximate solution, which can be leveraged to initialize the algorithm and improve the optimization process. Our numerical experiments show that the UQ model produces reliable estimates of the mean and STD of the approximate solution for the considered class of deep learning-based BSDE schemes. The estimated STD captures multiple sources of uncertainty, demonstrating its effectiveness in quantifying the uncertainty. Additionally, the model illustrates the improved performance when comparing different schemes based on the estimated STD values. Furthermore, it can identify hyperparameter values for which the scheme achieves good approximations.

Jonathan Klees, Tobias Riedlinger, Peter Stehr et al.

Safe artificial intelligence for perception tasks remains a major challenge, partly due to the lack of data with high-quality labels. Annotations themselves are subject to aleatoric and epistemic uncertainty, which is typically ignored during annotation and evaluation. While crowdsourcing enables collecting multiple annotations per image to estimate these uncertainties, this approach is impractical at scale due to the required annotation effort. We introduce a probabilistic label spreading method that provides reliable estimates of aleatoric and epistemic uncertainty of labels. Assuming label smoothness over the feature space, we propagate single annotations using a graph-based diffusion method. We prove that label spreading yields consistent probability estimators even when the number of annotations per data point converges to zero. We present and analyze a scalable implementation of our method. Experimental results indicate that, compared to baselines, our approach substantially reduces the annotation budget required to achieve a desired label quality on common image datasets and achieves a new state of the art on the Data-Centric Image Classification benchmark.

Jialong Wu, Yihan Wang, Matthias Rottmann

In autonomous driving, camera-radar fusion offers complementary sensing and low deployment cost. Existing methods perform fusion through input mixing, feature map mixing, or query-based feature sampling. We propose a new fusion paradigm, termed heterogeneous query interaction, and present ConFusion, a camera-radar 3D object detector. ConFusion combines image queries, radar queries, and learnable world queries distributed in 3D space to improve query initialization and object coverage. To encourage cross-type interaction among heterogeneous queries, we introduce heterogeneous query mixing (QMix), which performs dedicated cross-type attention after feature sampling to consolidate complementary object evidence. We further propose interactive query swap sampling (QSwap), which improves feature sampling by allowing related queries to exchange informative feature tokens under attention and geometric constraints. Experiments on the nuScenes dataset show that ConFusion achieves state-of-the-art performance, reaching 59.1 mAP and 65.6 NDS on the validation set, and 61.6 mAP and 67.9 NDS on the test set.

Edgar Heinert, Matthias Rottmann, Kira Maag et al.

Convolutional neural networks (CNNs) for image processing tend to focus on localized texture patterns, commonly referred to as texture bias. While most of the previous works in the literature focus on the task of image classification, we go beyond this and study the texture bias of CNNs in semantic segmentation. In this work, we propose to train CNNs on pre-processed images with less texture to reduce the texture bias. Therein, the challenge is to suppress image texture while preserving shape information. To this end, we utilize edge enhancing diffusion (EED), an anisotropic image diffusion method initially introduced for image compression, to create texture reduced duplicates of existing datasets. Extensive numerical studies are performed with both CNNs and vision transformer models trained on original data and EED-processed data from the Cityscapes dataset and the CARLA driving simulator. We observe strong texture-dependence of CNNs and moderate texture-dependence of transformers. Training CNNs on EED-processed images enables the models to become completely ignorant with respect to texture, demonstrating resilience with respect to texture re-introduction to any degree. Additionally we analyze the performance reduction in depth on a level of connected components in the semantic segmentation and study the influence of EED pre-processing on domain generalization as well as adversarial robustness.

Annika Mütze, Natalie Grabowsky, Edgar Heinert et al.

Recent research has investigated the shape and texture biases of pre-trained deep neural networks (DNNs) in image classification. Those works test how much a trained DNN relies on specific image cues like texture. The present study shifts the focus to understanding the cue influence during training, analyzing what DNNs can learn from shape, texture, and color cues in absence of the others; investigating their individual and combined influence on the learning success. We analyze these cue influences at multiple levels by decomposing datasets into cue-specific versions. Addressing semantic segmentation, we learn the given task from these reduced cue datasets, creating cue experts. Early fusion of cues is performed by constructing appropriate datasets. This is complemented by a late fusion of experts which allows us to study cue influence location-dependent on pixel level. Experiments on Cityscapes, PASCAL Context, and a synthetic CARLA dataset show that while no single cue dominates, the shape + color expert predominantly improves the prediction of small objects and border pixels. The cue performance order is consistent for the tested convolutional and transformer architecture, indicating similar cue extraction capabilities, although pre-trained transformers are said to be more biased towards shape than convolutional neural networks.

Mert Keser, Halil Ibrahim Orhan, Niki Amini-Naieni et al.

Deep neural networks (DNNs) remain challenged by distribution shifts in complex open-world domains like automated driving (AD): Robustness against yet unknown novel objects (semantic shift) or styles like lighting conditions (covariate shift) cannot be guaranteed. Hence, reliable operation-time monitors for identification of out-of-training-data-distribution (OOD) scenarios are imperative. Current approaches for OOD classification are untested for complex domains like AD, are limited in the kinds of shifts they detect, or even require supervision with OOD samples. To prepare for unanticipated shifts, we instead establish a framework around a principled, unsupervised and model-agnostic method that unifies detection of semantic and covariate shifts: Find a full model of the training data's feature distribution, to then use its density at new points as in-distribution (ID) score. To implement this, we propose to combine Vision Foundation Models (VFMs) as feature extractors with density modeling techniques. Through a comprehensive benchmark of 4 VFMs with different backbone architectures and 5 density-modeling techniques against established baselines, we provide the first systematic evaluation of OOD classification capabilities of VFMs across diverse conditions. A comparison with state-of-the-art binary OOD classification methods reveals that VFM embeddings with density estimation outperform existing approaches in identifying OOD inputs. Additionally, we show that our method detects high-risk inputs likely to cause errors in downstream tasks, thereby improving overall performance. Overall, VFMs, when coupled with robust density modeling techniques, are promising to realize model-agnostic, unsupervised, reliable safety monitors in complex vision tasks

Sarina Penquitt, Jonathan Klees, Rinor Cakaj et al.

Object detection has advanced rapidly in recent years, driven by increasingly large and diverse datasets. However, label errors, defined as missing labels, incorrect classification or inaccurate localization, often compromise the quality of these datasets. This can have a significant impact on the outcomes of training and benchmark evaluations. Although several methods now exist for detecting label errors in object detection datasets, they are typically validated only on synthetic benchmarks or limited manual inspection. How to correct such errors systemically and at scale therefore remains an open problem. We introduce a semi-automated framework for label-error correction called REC$\checkmark$D (Rechecked). Building on existing detectors, the framework pairs their error proposals with lightweight, crowd-sourced microtasks. These tasks enable multiple annotators to independently verify each candidate bounding box, and their responses are aggregated to estimate ambiguity and improve label quality. To demonstrate the effectiveness of REC$\checkmark$D, we apply it to the class pedestrian in the KITTI dataset. Our crowdsourced review yields high-quality corrected annotations, which indicate a rate of at least 24% of missing and inaccurate annotations in original annotations. This validated set will be released as a new real-world benchmark for label error detection and correction. We show that current label error detection methods, when combined with our correction framework, can recover hundreds of errors in the time it would take a human to annotate bounding boxes from scratch. However, even the best methods still miss up to 66% of the true errors and with low quality labels introduce more errors than they find. This highlights the urgent need for further research, now enabled by our released benchmark.

Ben Hamscher, Edgar Heinert, Annika Mütze et al.

Recent research has investigated the shape and texture biases of deep neural networks (DNNs) in image classification which influence their generalization capabilities and robustness. It has been shown that, in comparison to regular DNN training, training with stylized images reduces texture biases in image classification and improves robustness with respect to image corruptions. In an effort to advance this line of research, we examine whether style transfer can likewise deliver these two effects in semantic segmentation. To this end, we perform style transfer with style varying across artificial image areas. Those random areas are formed by a chosen number of Voronoi cells. The resulting style-transferred data is then used to train semantic segmentation DNNs with the objective of reducing their dependence on texture cues while enhancing their reliance on shape-based features. In our experiments, it turns out that in semantic segmentation, style transfer augmentation reduces texture bias and strongly increases robustness with respect to common image corruptions as well as adversarial attacks. These observations hold for convolutional neural networks and transformer architectures on the Cityscapes dataset as well as on PASCAL Context, showing the generality of the proposed method.

Martha Teiko Teye, Ori Maoz, Matthias Rottmann

Multi-object tracking from LiDAR point clouds presents unique challenges due to the sparse and irregular nature of the data, compounded by the need for temporal coherence across frames. Traditional tracking systems often rely on hand-crafted features and motion models, which can struggle to maintain consistent object identities in crowded or fast-moving scenes. We present a lidar-based two-staged DETR inspired transformer; a smoother and tracker. The smoother stage refines lidar object detections, from any off-the-shelf detector, across a moving temporal window. The tracker stage uses a DETR-based attention block to maintain tracks across time by associating tracked objects with the refined detections using the point cloud as context. The model is trained on the datasets nuScenes and KITTI in both online and offline (forward peeking) modes demonstrating strong performance across metrics such as ID-switch and multiple object tracking accuracy (MOTA). The numerical results indicate that the online mode outperforms the lidar-only baseline and SOTA models on the nuScenes dataset, with an aMOTA of 0.724 and an aMOTP of 0.475, while the offline mode provides an additional 3 pp aMOTP.

Antonia van Betteray, Matthias Rottmann, Karsten Kahl

The structural analogies of ResNets and Multigrid (MG) methods such as common building blocks like convolutions and poolings where already pointed out by He et al.\ in 2016. Multigrid methods are used in the context of scientific computing for solving large sparse linear systems arising from partial differential equations. MG methods particularly rely on two main concepts: smoothing and residual restriction / coarsening. Exploiting these analogies, He and Xu developed the MgNet framework, which integrates MG schemes into the design of ResNets. In this work, we introduce a novel neural network building block inspired by polynomial smoothers from MG theory. Our polynomial block from an MG perspective naturally extends the MgNet framework to Poly-Mgnet and at the same time reduces the number of weights in MgNet. We present a comprehensive study of our polynomial block, analyzing the choice of initial coefficients, the polynomial degree, the placement of activation functions, as well as of batch normalizations. Our results demonstrate that constructing (quadratic) polynomial building blocks based on real and imaginary polynomial roots enhances Poly-MgNet's capacity in terms of accuracy. Furthermore, our approach achieves an improved trade-off of model accuracy and number of weights compared to ResNet as well as compared to specific configurations of MgNet.

Edgar Heinert, Thomas Gottwald, Annika Mütze et al.

Previous works studied how deep neural networks (DNNs) perceive image content in terms of their biases towards different image cues, such as texture and shape. Previous methods to measure shape and texture biases are typically style-transfer-based and limited to DNNs for image classification. In this work, we provide a new evaluation procedure consisting of 1) a cue-decomposition method that comprises two AI-free data pre-processing methods extracting shape and texture cues, respectively, and 2) a novel cue-decomposition shape bias evaluation metric that leverages the cue-decomposition data. For application purposes we introduce a corresponding cue-decomposition robustness metric that allows for the estimation of the robustness of a DNN w.r.t. image corruptions. In our numerical experiments, our findings for biases in image classification DNNs align with those of previous evaluation metrics. However, our cue-decomposition robustness metric shows superior results in terms of estimating the robustness of DNNs. Furthermore, our results for DNNs on the semantic segmentation datasets Cityscapes and ADE20k for the first time shed light into the biases of semantic segmentation DNNs.

Martha Teiko Teye, Ori Maoz, Matthias Rottmann

We propose FutrTrack, a modular camera-LiDAR multi-object tracking framework that builds on existing 3D detectors by introducing a transformer-based smoother and a fusion-driven tracker. Inspired by query-based tracking frameworks, FutrTrack employs a multimodal two-stage transformer refinement and tracking pipeline. Our fusion tracker integrates bounding boxes with multimodal bird's-eye-view (BEV) fusion features from multiple cameras and LiDAR without the need for an explicit motion model. The tracker assigns and propagates identities across frames, leveraging both geometric and semantic cues for robust re-identification under occlusion and viewpoint changes. Prior to tracking, we refine sequences of bounding boxes with a temporal smoother over a moving window to refine trajectories, reduce jitter, and improve spatial consistency. Evaluated on nuScenes and KITTI, FutrTrack demonstrates that query-based transformer tracking methods benefit significantly from multimodal sensor features compared with previous single-sensor approaches. With an aMOTA of 74.7 on the nuScenes test set, FutrTrack achieves strong performance on 3D MOT benchmarks, reducing identity switches while maintaining competitive accuracy. Our approach provides an efficient framework for improving transformer-based trackers to compete with other neural-network-based methods even with limited data and without pretraining.

Sarina Penquitt, Tobias Riedlinger, Timo Heller et al.

Recently, detection of label errors and improvement of label quality in datasets for supervised learning tasks has become an increasingly important goal in both research and industry. The consequences of incorrectly annotated data include reduced model performance, biased benchmark results, and lower overall accuracy. Current state-of-the-art label error detection methods often focus on a single computer vision task and, consequently, a specific type of dataset, containing, for example, either bounding boxes or pixel-wise annotations. Furthermore, previous methods are not learning-based. In this work, we overcome this research gap. We present a unified method for detecting label errors in object detection, semantic segmentation, and instance segmentation datasets. In a nutshell, our approach - learning to detect label errors by making them - works as follows: we inject different kinds of label errors into the ground truth. Then, the detection of label errors, across all mentioned primary tasks, is framed as an instance segmentation problem based on a composite input. In our experiments, we compare the label error detection performance of our method with various baselines and state-of-the-art approaches of each task's domain on simulated label errors across multiple tasks, datasets, and base models. This is complemented by a generalization study on real-world label errors. Additionally, we release 459 real label errors identified in the Cityscapes dataset and provide a benchmark for real label error detection in Cityscapes.

Thomas Gottwald, Edgar Heinert, Matthias Rottmann

In this work, we present a novel method for uncertainty estimation (UE) in Gaussian Splatting. UE is crucial for using Gaussian Splatting in critical applications such as robotics and medicine. Previous methods typically estimate the variance of Gaussian primitives and use the rendering process to obtain pixel-wise uncertainties. Our method establishes primitive representations of error and visibility of trainings views, which carries meaningful uncertainty information. This representation is obtained by projection of training error and visibility onto the primitives. Uncertainties of novel views are obtained by rendering the primitive representations of uncertainty for those novel views, yielding uncertainty feature maps. To aggregate these uncertainty feature maps of novel views, we perform a pixel-wise regression on holdout data. In our experiments, we analyze the different components of our method, investigating various combinations of uncertainty feature maps and regression models. Furthermore, we considered the effect of separating splatting into foreground and background. Our UEs show high correlations to true errors, outperforming state-of-the-art methods, especially on foreground objects. The trained regression models show generalization capabilities to new scenes, allowing uncertainty estimation without the need for holdout data.

Annika Mütze, Sadia Ilyas, Christian Dörpelkus et al.

Open-vocabulary object detectors such as Grounding DINO are trained on vast and diverse data, achieving remarkable performance on challenging datasets. Due to that, it is unclear where to find their limitations, which is of major concern when using in safety-critical applications. Real-world data does not provide sufficient control, required for a rigorous evaluation of model generalization. In contrast, synthetically generated data allows to systematically explore the boundaries of model competence/generalization. In this work, we address two research questions: 1) Can we challenge open-vocabulary object detectors with generated image content? 2) Can we find systematic failure modes of those models? To address these questions, we design two automated pipelines using stable diffusion to inpaint unusual objects with high diversity in semantics, by sampling multiple substantives from WordNet and ChatGPT. On the synthetically generated data, we evaluate and compare multiple open-vocabulary object detectors as well as a classical object detector. The synthetic data is derived from two real-world datasets, namely LostAndFound, a challenging out-of-distribution (OOD) detection benchmark, and the NuImages dataset. Our results indicate that inpainting can challenge open-vocabulary object detectors in terms of overlooking objects. Additionally, we find a strong dependence of open-vocabulary models on object location, rather than on object semantics. This provides a systematic approach to challenge open-vocabulary models and gives valuable insights on how data could be acquired to effectively improve these models.

Antonia van Betteray, Matthias Rottmann, Karsten Kahl

The singular values of convolutional mappings encode interesting spectral properties, which can be used, e.g., to improve generalization and robustness of convolutional neural networks as well as to facilitate model compression. However, the computation of singular values is typically very resource-intensive. The naive approach involves unrolling the convolutional mapping along the input and channel dimensions into a large and sparse two-dimensional matrix, making the exact calculation of all singular values infeasible due to hardware limitations. In particular, this is true for matrices that represent convolutional mappings with large inputs and a high number of channels. Existing efficient methods leverage the Fast Fourier transformation (FFT) to transform convolutional mappings into the frequency domain, enabling the computation of singular values for matrices representing convolutions with larger input and channel dimensions. For a constant number of channels in a given convolution, an FFT can compute N singular values in O(N log N) complexity. In this work, we propose an approach of complexity O(N) based on local Fourier analysis, which additionally exploits the shift invariance of convolutional operators. We provide a theoretical analysis of our algorithm's runtime and validate its efficiency through numerical experiments. Our results demonstrate that our proposed method is scalable and offers a practical solution to calculate the entire set of singular values - along with the corresponding singular vectors if needed - for high-dimensional convolutional mappings.

Vignesh Gopinathan, Urs Zimmermann, Michael Arnold et al.

Video captioning models have seen notable advancements in recent years, especially with regard to their ability to capture temporal information. While many research efforts have focused on architectural advancements, such as temporal attention mechanisms, there remains a notable gap in understanding how models capture and utilize temporal semantics for effective temporal feature extraction, especially in the context of Advanced Driver Assistance Systems. We propose an automated LiDAR-based captioning procedure that focuses on the temporal dynamics of traffic participants. Our approach uses a rule-based system to extract essential details such as lane position and relative motion from object tracks, followed by a template-based caption generation. Our findings show that training SwinBERT, a video captioning model, using only front camera images and supervised with our template-based captions, specifically designed to encapsulate fine-grained temporal behavior, leads to improved temporal understanding consistently across three datasets. In conclusion, our results clearly demonstrate that integrating LiDAR-based caption supervision significantly enhances temporal understanding, effectively addressing and reducing the inherent visual/static biases prevalent in current state-of-the-art model architectures.

Gesina Schwalbe, Georgii Mikriukov, Edgar Heinert et al.

The thriving research field of concept-based explainable artificial intelligence (C-XAI) investigates how human-interpretable semantic concepts embed in the latent spaces of deep neural networks (DNNs). Post-hoc approaches therein use a set of examples to specify a concept, and determine its embeddings in DNN latent space using data driven techniques. This proved useful to uncover biases between different target (foreground or concept) classes. However, given that the background is mostly uncontrolled during training, an important question has been left unattended so far: Are/to what extent are state-of-the-art, data-driven post-hoc C-XAI approaches themselves prone to biases with respect to their backgrounds? E.g., wild animals mostly occur against vegetation backgrounds, and they seldom appear on roads. Even simple and robust C-XAI methods might abuse this shortcut for enhanced performance. A dangerous performance degradation of the concept-corner cases of animals on the road could thus remain undiscovered. This work validates and thoroughly confirms that established Net2Vec-based concept segmentation techniques frequently capture background biases, including alarming ones, such as underperformance on road scenes. For the analysis, we compare 3 established techniques from the domain of background randomization on >50 concepts from 2 datasets, and 7 diverse DNN architectures. Our results indicate that even low-cost setups can provide both valuable insight and improved background robustness.

Jialong Wu, Marco Braun, Dominic Spata et al.

Scene flow provides crucial motion information for autonomous driving. Recent LiDAR scene flow models utilize the rigid-motion assumption at the instance level, assuming objects are rigid bodies. However, these instance-level methods are not suitable for sparse radar point clouds. In this work, we present a novel Traffic-Aware Radar Scene-Flow (TARS) estimation method, which utilizes motion rigidity at the traffic level. To address the challenges in radar scene flow, we perform object detection and scene flow jointly and boost the latter. We incorporate the feature map from the object detector, trained with detection losses, to make radar scene flow aware of the environment and road users. From this, we construct a Traffic Vector Field (TVF) in the feature space to achieve holistic traffic-level scene understanding in our scene flow branch. When estimating the scene flow, we consider both point-level motion cues from point neighbors and traffic-level consistency of rigid motion within the space. TARS outperforms the state of the art on a proprietary dataset and the View-of-Delft dataset, improving the benchmarks by 23% and 15%, respectively.

Natalie Grabowsky, Annika Mütze, Joshua Wendland et al.

Agents in real-world scenarios like automated driving deal with uncertainty in their environment, in particular due to perceptual uncertainty. Although, reinforcement learning is dedicated to autonomous decision-making under uncertainty these algorithms are typically not informed about the uncertainty currently contained in their environment. On the other hand, uncertainty estimation for perception itself is typically directly evaluated in the perception domain, e.g., in terms of false positive detection rates or calibration errors based on camera images. Its use for deciding on goal-oriented actions remains largely unstudied. In this paper, we investigate how an agent's behavior is influenced by an uncertain perception and how this behavior changes if information about this uncertainty is available. Therefore, we consider a proxy task, where the agent is rewarded for driving a route as fast as possible without colliding with other road users. For controlled experiments, we introduce uncertainty in the observation space by perturbing the perception of the given agent while informing the latter. Our experiments show that an unreliable observation space modeled by a perturbed perception leads to a defensive driving behavior of the agent. Furthermore, when adding the information about the current uncertainty directly to the observation space, the agent adapts to the specific situation and in general accomplishes its task faster while, at the same time, accounting for risks.

Alexey Nekrasov, Rui Zhou, Miriam Ackermann et al.

Safe navigation of self-driving cars and robots requires a precise understanding of their environment. Training data for perception systems cannot cover the wide variety of objects that may appear during deployment. Thus, reliable identification of unknown objects, such as wild animals and untypical obstacles, is critical due to their potential to cause serious accidents. Significant progress in semantic segmentation of anomalies has been facilitated by the availability of out-of-distribution (OOD) benchmarks. However, a comprehensive understanding of scene dynamics requires the segmentation of individual objects, and thus the segmentation of instances is essential. Development in this area has been lagging, largely due to the lack of dedicated benchmarks. The situation is similar in object detection. While there is interest in detecting and potentially tracking every anomalous object, the availability of dedicated benchmarks is clearly limited. To address this gap, this work extends some commonly used anomaly segmentation benchmarks to include the instance segmentation and object detection tasks. Our evaluation of anomaly instance segmentation and object detection methods shows that both of these challenges remain unsolved problems. We provide a competition and benchmark website under https://vision.rwth-aachen.de/oodis

Jialong Wu, Mirko Meuter, Markus Schoeler et al.

Radar-based perception has gained increasing attention in autonomous driving, yet the inherent sparsity of radars poses challenges. Radar raw data often contains excessive noise, whereas radar point clouds retain only limited information. In this work, we holistically treat the sparse nature of radar data by introducing an adaptive subsampling method together with a tailored network architecture that exploits the sparsity patterns to discover global and local dependencies in the radar signal. Our subsampling module selects a subset of pixels from range-doppler (RD) spectra that contribute most to the downstream perception tasks. To improve the feature extraction on sparse subsampled data, we propose a new way of applying graph neural networks on radar data and design a novel two-branch backbone to capture both global and local neighbor information. An attentive fusion module is applied to combine features from both branches. Experiments on the RADIal dataset show that our SparseRadNet exceeds state-of-the-art (SOTA) performance in object detection and achieves close to SOTA accuracy in freespace segmentation, meanwhile using sparse subsampled input data.

Robin Chan, Svenja Uhlemeyer, Matthias Rottmann et al.

Semantic segmentation is a crucial component for perception in automated driving. Deep neural networks (DNNs) are commonly used for this task and they are usually trained on a closed set of object classes appearing in a closed operational domain. However, this is in contrast to the open world assumption in automated driving that DNNs are deployed to. Therefore, DNNs necessarily face data that they have never encountered previously, also known as anomalies, which are extremely safety-critical to properly cope with. In this work, we first give an overview about anomalies from an information-theoretic perspective. Next, we review research in detecting semantically unknown objects in semantic segmentation. We demonstrate that training for high entropy responses on anomalous objects outperforms other recent methods, which is in line with our theoretical findings. Moreover, we examine a method to assess the occurrence frequency of anomalies in order to select anomaly types to include into a model's set of semantic categories. We demonstrate that these anomalies can then be learned in an unsupervised fashion, which is particularly suitable in online applications based on deep learning.

Philipp Oberdiek, Gernot A. Fink, Matthias Rottmann

We present an approach to quantifying both aleatoric and epistemic uncertainty for deep neural networks in image classification, based on generative adversarial networks (GANs). While most works in the literature that use GANs to generate out-of-distribution (OoD) examples only focus on the evaluation of OoD detection, we present a GAN based approach to learn a classifier that produces proper uncertainties for OoD examples as well as for false positives (FPs). Instead of shielding the entire in-distribution data with GAN generated OoD examples which is state-of-the-art, we shield each class separately with out-of-class examples generated by a conditional GAN and complement this with a one-vs-all image classifier. In our experiments, in particular on CIFAR10, CIFAR100 and Tiny ImageNet, we improve over the OoD detection and FP detection performance of state-of-the-art GAN-training based classifiers. Furthermore, we also find that the generated GAN examples do not significantly affect the calibration error of our classifier and result in a significant gain in model accuracy.

Svenja Uhlemeyer, Matthias Rottmann, Hanno Gottschalk

For the semantic segmentation of images, state-of-the-art deep neural networks (DNNs) achieve high segmentation accuracy if that task is restricted to a closed set of classes. However, as of now DNNs have limited ability to operate in an open world, where they are tasked to identify pixels belonging to unknown objects and eventually to learn novel classes, incrementally. Humans have the capability to say: I don't know what that is, but I've already seen something like that. Therefore, it is desirable to perform such an incremental learning task in an unsupervised fashion. We introduce a method where unknown objects are clustered based on visual similarity. Those clusters are utilized to define new classes and serve as training data for unsupervised incremental learning. More precisely, the connected components of a predicted semantic segmentation are assessed by a segmentation quality estimate. connected components with a low estimated prediction quality are candidates for a subsequent clustering. Additionally, the component-wise quality assessment allows for obtaining predicted segmentation masks for the image regions potentially containing unknown objects. The respective pixels of such masks are pseudo-labeled and afterwards used for re-training the DNN, i.e., without the use of ground truth generated by humans. In our experiments we demonstrate that, without access to ground truth and even with few data, a DNN's class space can be extended by a novel class, achieving considerable segmentation accuracy.

Hanno Gottschalk, Matthias Rottmann, Maida Saltagic

While automated driving is often advertised with better-than-human driving performance, this work reviews that it is nearly impossible to provide direct statistical evidence on the system level that this is actually the case. The amount of labeled data needed would exceed dimensions of present day technical and economical capabilities. A commonly used strategy therefore is the use of redundancy along with the proof of sufficient subsystems' performances. As it is known, this strategy is efficient especially for the case of subsystems operating independently, i.e. the occurrence of errors is independent in a statistical sense. Here, we give some first considerations and experimental evidence that this strategy is not a free ride as the errors of neural networks fulfilling the same computer vision task, at least for some cases, show correlated occurrences of errors. This remains true, if training data, architecture, and training are kept separate or independence is trained using special loss functions. Using data from different sensors (realized by up to five 2D projections of the 3D MNIST data set) in our experiments is more efficiently reducing correlations, however not to an extent that is realizing the potential of reduction of testing data that can be obtained for redundant and statistically independent subsystems.

Pascal Colling, Matthias Rottmann, Lutz Roese-Koerner et al.

We present a novel post-processing tool for semantic segmentation of LiDAR point cloud data, called LidarMetaSeg, which estimates the prediction quality segmentwise. For this purpose we compute dispersion measures based on network probability outputs as well as feature measures based on point cloud input features and aggregate them on segment level. These aggregated measures are used to train a meta classification model to predict whether a predicted segment is a false positive or not and a meta regression model to predict the segmentwise intersection over union. Both models can then be applied to semantic segmentation inferences without knowing the ground truth. In our experiments we use different LiDAR segmentation models and datasets and analyze the power of our method. We show that our results outperform other standard approaches.

Claudia Drygala, Matthias Rottmann, Hanno Gottschalk et al.

To ensure safety in automated driving, the correct perception of the situation inside the car is as important as its environment. Thus, seat occupancy detection and classification of detected instances play an important role in interior sensing. By the knowledge of the seat occupancy status, it is possible to, e.g., automate the airbag deployment control. Furthermore, the presence of a driver, which is necessary for partially automated driving cars at the automation levels two to four can be verified. In this work, we compare different statistical methods from the field of image segmentation to approach the problem of background-foreground segmentation in camera based interior sensing. In the recent years, several methods based on different techniques have been developed and applied to images or videos from different applications. The peculiarity of the given scenarios of interior sensing is, that the foreground instances and the background both contain static as well as dynamic elements. In data considered in this work, even the camera position is not completely fixed. We review and benchmark three different methods ranging, i.e., Gaussian Mixture Models (GMM), Morphological Snakes and a deep neural network, namely a Mask R-CNN. In particular, the limitations of the classical methods, GMM and Morphological Snakes, for interior sensing are shown. Furthermore, it turns, that it is possible to overcome these limitations by deep learning, e.g.\ using a Mask R-CNN. Although only a small amount of ground truth data was available for training, we enabled the Mask R-CNN to produce high quality background-foreground masks via transfer learning. Moreover, we demonstrate that certain augmentation as well as pre- and post-processing methods further enhance the performance of the investigated methods.

Tobias Riedlinger, Matthias Rottmann, Marius Schubert et al.

The vast majority of uncertainty quantification methods for deep object detectors such as variational inference are based on the network output. Here, we study gradient-based epistemic uncertainty metrics for deep object detectors to obtain reliable confidence estimates. We show that they contain predictive information and that they capture information orthogonal to that of common, output-based uncertainty estimation methods like Monte-Carlo dropout and deep ensembles. To this end, we use meta classification and meta regression to produce confidence estimates using gradient metrics and other baselines for uncertainty quantification which are in principle applicable to any object detection architecture. Specifically, we employ false positive detection and prediction of localization quality to investigate uncertainty content of our metrics and compute the calibration errors of meta classifiers. Moreover, we use them as a post-processing filter mechanism to the object detection pipeline and compare object detection performance. Our results show that gradient-based uncertainty is itself on par with output-based methods across different detectors and datasets. More significantly, combined meta classifiers based on gradient and output-based metrics outperform the standalone models. Based on this result, we conclude that gradient uncertainty adds orthogonal information to output-based methods. This suggests that variational inference may be supplemented by gradient-based uncertainty to obtain improved confidence measures, contributing to down-stream applications of deep object detectors and improving their probabilistic reliability.

Julia Rosenzweig, Eduardo Brito, Hans-Ulrich Kobialka et al.

Many machine learning applications can benefit from simulated data for systematic validation - in particular if real-life data is difficult to obtain or annotate. However, since simulations are prone to domain shift w.r.t. real-life data, it is crucial to verify the transferability of the obtained results. We propose a novel framework consisting of a generative label-to-image synthesis model together with different transferability measures to inspect to what extent we can transfer testing results of semantic segmentation models from synthetic data to equivalent real-life data. With slight modifications, our approach is extendable to, e.g., general multi-class classification tasks. Grounded on the transferability analysis, our approach additionally allows for extensive testing by incorporating controlled simulations. We validate our approach empirically on a semantic segmentation task on driving scenes. Transferability is tested using correlation analysis of IoU and a learned discriminator. Although the latter can distinguish between real-life and synthetic tests, in the former we observe surprisingly strong correlations of 0.7 for both cars and pedestrians.

Robin Chan, Krzysztof Lis, Svenja Uhlemeyer et al.

State-of-the-art semantic or instance segmentation deep neural networks (DNNs) are usually trained on a closed set of semantic classes. As such, they are ill-equipped to handle previously-unseen objects. However, detecting and localizing such objects is crucial for safety-critical applications such as perception for automated driving, especially if they appear on the road ahead. While some methods have tackled the tasks of anomalous or out-of-distribution object segmentation, progress remains slow, in large part due to the lack of solid benchmarks; existing datasets either consist of synthetic data, or suffer from label inconsistencies. In this paper, we bridge this gap by introducing the "SegmentMeIfYouCan" benchmark. Our benchmark addresses two tasks: Anomalous object segmentation, which considers any previously-unseen object category; and road obstacle segmentation, which focuses on any object on the road, may it be known or unknown. We provide two corresponding datasets together with a test suite performing an in-depth method analysis, considering both established pixel-wise performance metrics and recent component-wise ones, which are insensitive to object sizes. We empirically evaluate multiple state-of-the-art baseline methods, including several models specifically designed for anomaly / obstacle segmentation, on our datasets and on public ones, using our test suite. The anomaly and obstacle segmentation results show that our datasets contribute to the diversity and difficulty of both data landscapes.