Bodo Rosenhahn

Timo Kaiser, Lukas Ehmann, Christoph Reinders et al.

Optimizing neural networks with noisy labels is a challenging task, especially if the label set contains real-world noise. Networks tend to generalize to reasonable patterns in the early training stages and overfit to specific details of noisy samples in the latter ones. We introduce Blind Knowledge Distillation - a novel teacher-student approach for learning with noisy labels by masking the ground truth related teacher output to filter out potentially corrupted knowledge and to estimate the tipping point from generalizing to overfitting. Based on this, we enable the estimation of noise in the training data with Otsus algorithm. With this estimation, we train the network with a modified weighted cross-entropy loss function. We show in our experiments that Blind Knowledge Distillation detects overfitting effectively during training and improves the detection of clean and noisy labels on the recently published CIFAR-N dataset. Code is available at GitHub.

Timo Kaiser, Vladimir Ulman, Bodo Rosenhahn

The evaluation of cell tracking results steers the development of tracking methods, significantly impacting biomedical research. This is quantitatively achieved by means of evaluation metrics. Unfortunately, current metrics favor local correctness and weakly reward global coherence, impeding high-level biological analysis. To also foster global coherence, we propose the CHOTA metric (Cell-specific Higher Order Tracking Accuracy) which unifies the evaluation of all relevant aspects of cell tracking: cell detections and local associations, global coherence, and lineage tracking. We achieve this by introducing a new definition of the term 'trajectory' that includes the entire cell lineage and by including this into the well-established HOTA metric from general multiple object tracking. Furthermore, we provide a detailed survey of contemporary cell tracking metrics to compare our novel CHOTA metric and to show its advantages. All metrics are extensively evaluated on state-of-the-art real-data cell tracking results and synthetic results that simulate specific tracking errors. We show that CHOTA is sensitive to all tracking errors and gives a good indication of the biologically relevant capability of a method to reconstruct the full lineage of cells. It introduces a robust and comprehensive alternative to the currently used metrics in cell tracking. Python code is available at https://github.com/CellTrackingChallenge/py-ctcmetrics .

Marco Rudolph, Tom Wehrbein, Bodo Rosenhahn et al.

Industrial defect detection is commonly addressed with anomaly detection (AD) methods where no or only incomplete data of potentially occurring defects is available. This work discovers previously unknown problems of student-teacher approaches for AD and proposes a solution, where two neural networks are trained to produce the same output for the defect-free training examples. The core assumption of student-teacher networks is that the distance between the outputs of both networks is larger for anomalies since they are absent in training. However, previous methods suffer from the similarity of student and teacher architecture, such that the distance is undesirably small for anomalies. For this reason, we propose asymmetric student-teacher networks (AST). We train a normalizing flow for density estimation as a teacher and a conventional feed-forward network as a student to trigger large distances for anomalies: The bijectivity of the normalizing flow enforces a divergence of teacher outputs for anomalies compared to normal data. Outside the training distribution the student cannot imitate this divergence due to its fundamentally different architecture. Our AST network compensates for wrongly estimated likelihoods by a normalizing flow, which was alternatively used for anomaly detection in previous work. We show that our method produces state-of-the-art results on the two currently most relevant defect detection datasets MVTec AD and MVTec 3D-AD regarding image-level anomaly detection on RGB and 3D data.

Patrick Glandorf, Thomas Norrenbrock, Bodo Rosenhahn

Vision Transformers (ViTs) have demonstrated state-ofthe-art performance in several benchmarks, yet their high computational costs hinders their practical deployment. Patch Pruning offers significant savings, but existing approaches restrict token reduction to deeper layers, leaving early-stage compression unexplored. This limits their potential for holistic efficiency. In this work, we present a novel Video Patch Pruning framework (VPP) that integrates temporal prior knowledge to enable efficient sparsity within early ViT layers. Our approach is motivated by the observation that prior features extracted from deeper layers exhibit strong foreground selectivity. Therefore we propose a fully differentiable module for temporal mapping to accurately select the most relevant patches in early network stages. Notably, the proposed method enables a patch reduction of up to 60% in dense prediction tasks, exceeding the capabilities of conventional image-based patch pruning, which typically operate around a 30% patch sparsity. VPP excels the high-sparsity regime, sustaining remarkable performance even when patch usage is reduced below 55%. Specifically, it preserves stable results with a maximal performance drop of 0.6% on the Youtube-VIS 2021 dataset.

Melissa Schween, Mathis Kruse, Bodo Rosenhahn

We propose Bijective Universal Scene-Specific Anomalous Relationship Detection (BUSSARD), a normalizing flow-based model for detecting anomalous relations in scene graphs, generated from images. Our work follows a multimodal approach, embedding object and relationship tokens from scene graphs with a language model to leverage semantic knowledge from the real world. A normalizing flow model is used to learn bijective transformations that map object-relation-object triplets from scene graphs to a simple base distribution (typically Gaussian), allowing anomaly detection through likelihood estimation. We evaluate our approach on the SARD dataset containing office and dining room scenes. Our method achieves around 10% better AUROC results compared to the current state-of-the-art model, while simultaneously being five times faster. Through ablation studies, we demonstrate superior robustness and universality, particularly regarding the use of synonyms, with our model maintaining stable performance while the baseline shows 17.5% deviation. This work demonstrates the strong potential of learning-based methods for relationship anomaly detection in scene graphs. Our code is available at https://github.com/mschween/BUSSARD .

Andrea Avogaro, Federico Cunico, Bodo Rosenhahn et al.

Markerless Human Pose Estimation (HPE) proved its potential to support decision making and assessment in many fields of application. HPE is often preferred to traditional marker-based Motion Capture systems due to the ease of setup, portability, and affordable cost of the technology. However, the exploitation of HPE in biomedical applications is still under investigation. This review aims to provide an overview of current biomedical applications of HPE. In this paper, we examine the main features of HPE approaches and discuss whether or not those features are of interest to biomedical applications. We also identify those areas where HPE is already in use and present peculiarities and trends followed by researchers and practitioners. We include here 25 approaches to HPE and more than 40 studies of HPE applied to motor development assessment, neuromuscolar rehabilitation, and gait & posture analysis. We conclude that markerless HPE offers great potential for extending diagnosis and rehabilitation outside hospitals and clinics, toward the paradigm of remote medical care.

Vasileios Iosifidis, Symeon Papadopoulos, Bodo Rosenhahn et al.

Class imbalance poses a major challenge for machine learning as most supervised learning models might exhibit bias towards the majority class and under-perform in the minority class. Cost-sensitive learning tackles this problem by treating the classes differently, formulated typically via a user-defined fixed misclassification cost matrix provided as input to the learner. Such parameter tuning is a challenging task that requires domain knowledge and moreover, wrong adjustments might lead to overall predictive performance deterioration. In this work, we propose a novel cost-sensitive boosting approach for imbalanced data that dynamically adjusts the misclassification costs over the boosting rounds in response to model's performance instead of using a fixed misclassification cost matrix. Our method, called AdaCC, is parameter-free as it relies on the cumulative behavior of the boosting model in order to adjust the misclassification costs for the next boosting round and comes with theoretical guarantees regarding the training error. Experiments on 27 real-world datasets from different domains with high class imbalance demonstrate the superiority of our method over 12 state-of-the-art cost-sensitive boosting approaches exhibiting consistent improvements in different measures, for instance, in the range of [0.3%-28.56%] for AUC, [3.4%-21.4%] for balanced accuracy, [4.8%-45%] for gmean and [7.4%-85.5%] for recall.

Yuren Cong, Martin Renqiang Min, Li Erran Li et al.

Despite the recent impressive breakthroughs in text-to-image generation, generative models have difficulty in capturing the data distribution of underrepresented attribute compositions while over-memorizing overrepresented attribute compositions, which raises public concerns about their robustness and fairness. To tackle this challenge, we propose ACTIG, an attribute-centric compositional text-to-image generation framework. We present an attribute-centric feature augmentation and a novel image-free training scheme, which greatly improves model's ability to generate images with underrepresented attributes. We further propose an attribute-centric contrastive loss to avoid overfitting to overrepresented attribute compositions. We validate our framework on the CelebA-HQ and CUB datasets. Extensive experiments show that the compositional generalization of ACTIG is outstanding, and our framework outperforms previous works in terms of image quality and text-image consistency.

Jan Thieß Brockmann, Marco Rudolph, Bodo Rosenhahn et al.

During the operation of industrial robots, unusual events may endanger the safety of humans and the quality of production. When collecting data to detect such cases, it is not ensured that data from all potentially occurring errors is included as unforeseeable events may happen over time. Therefore, anomaly detection (AD) delivers a practical solution, using only normal data to learn to detect unusual events. We introduce a dataset that allows training and benchmarking of anomaly detection methods for robotic applications based on machine data which will be made publicly available to the research community. As a typical robot task the dataset includes a pick-and-place application which involves movement, actions of the end effector and interactions with the objects of the environment. Since several of the contained anomalies are not task-specific but general, evaluations on our dataset are transferable to other robotics applications as well. Additionally, we present MVT-Flow (multivariate time-series flow) as a new baseline method for anomaly detection: It relies on deep-learning-based density estimation with normalizing flows, tailored to the data domain by taking its structure into account for the architecture. Our evaluation shows that MVT-Flow outperforms baselines from previous work by a large margin of 6.2% in area under ROC.

Lutz M. K. Krause, Emily Manderfeld, Patricia Gnutt et al.

Biofouling is a major challenge for sustainable shipping, filter membranes, heat exchangers, and medical devices. The development of fouling-resistant coatings requires the evaluation of their effectiveness. Such an evaluation is usually based on the assessment of fouling progression after different exposure times to the target medium (e.g., salt water). The manual assessment of macrofouling requires expert knowledge about local fouling communities due to high variances in phenotypical appearance, has single-image sampling inaccuracies for certain species, and lacks spatial information. Here we present an approach for automatic image-based macrofouling analysis. We created a dataset with dense labels prepared from field panel images and propose a convolutional network (adapted U-Net) for the semantic segmentation of different macrofouling classes. The establishment of macrofouling localization allows for the generation of a successional model which enables the determination of direct surface attachment and in-depth epibiotic studies.

Thomas Norrenbrock, Marco Rudolph, Bodo Rosenhahn

Deep Neural Networks use thousands of mostly incomprehensible features to identify a single class, a decision no human can follow. We propose an interpretable sparse and low dimensional final decision layer in a deep neural network with measurable aspects of interpretability and demonstrate it on fine-grained image classification. We argue that a human can only understand the decision of a machine learning model, if the features are interpretable and only very few of them are used for a single decision. For that matter, the final layer has to be sparse and, to make interpreting the features feasible, low dimensional. We call a model with a Sparse Low-Dimensional Decision SLDD-Model. We show that a SLDD-Model is easier to interpret locally and globally than a dense high-dimensional decision layer while being able to maintain competitive accuracy. Additionally, we propose a loss function that improves a model's feature diversity and accuracy. Our more interpretable SLDD-Model only uses 5 out of just 50 features per class, while maintaining 97% to 100% of the accuracy on four common benchmark datasets compared to the baseline model with 2048 features.

Yuren Cong, Jinhui Yi, Bodo Rosenhahn et al.

As a natural extension of the image synthesis task, video synthesis has attracted a lot of interest recently. Many image synthesis works utilize class labels or text as guidance. However, neither labels nor text can provide explicit temporal guidance, such as when an action starts or ends. To overcome this limitation, we introduce semantic video scene graphs as input for video synthesis, as they represent the spatial and temporal relationships between objects in the scene. Since video scene graphs are usually temporally discrete annotations, we propose a video scene graph (VSG) encoder that not only encodes the existing video scene graphs but also predicts the graph representations for unlabeled frames. The VSG encoder is pre-trained with different contrastive multi-modal losses. A semantic scene graph-to-video synthesis framework (SSGVS), based on the pre-trained VSG encoder, VQ-VAE, and auto-regressive Transformer, is proposed to synthesize a video given an initial scene image and a non-fixed number of semantic scene graphs. We evaluate SSGVS and other state-of-the-art video synthesis models on the Action Genome dataset and demonstrate the positive significance of video scene graphs in video synthesis. The source code will be released.

Patrick Glandorf, Timo Kaiser, Bodo Rosenhahn

Sparse neural networks are a key factor in developing resource-efficient machine learning applications. We propose the novel and powerful sparse learning method Adaptive Regularized Training (ART) to compress dense into sparse networks. Instead of the commonly used binary mask during training to reduce the number of model weights, we inherently shrink weights close to zero in an iterative manner with increasing weight regularization. Our method compresses the pre-trained model knowledge into the weights of highest magnitude. Therefore, we introduce a novel regularization loss named HyperSparse that exploits the highest weights while conserving the ability of weight exploration. Extensive experiments on CIFAR and TinyImageNet show that our method leads to notable performance gains compared to other sparsification methods, especially in extremely high sparsity regimes up to 99.8 percent model sparsity. Additional investigations provide new insights into the patterns that are encoded in weights with high magnitudes.

Timo Kaiser, Christoph Reinders, Bodo Rosenhahn

Label noise and ambiguities between similar classes are challenging problems in developing new models and annotating new data for semantic segmentation. In this paper, we propose Compensation Learning in Semantic Segmentation, a framework to identify and compensate ambiguities as well as label noise. More specifically, we add a ground truth depending and globally learned bias to the classification logits and introduce a novel uncertainty branch for neural networks to induce the compensation bias only to relevant regions. Our method is employed into state-of-the-art segmentation frameworks and several experiments demonstrate that our proposed compensation learns inter-class relations that allow global identification of challenging ambiguities as well as the exact localization of subsequent label noise. Additionally, it enlarges robustness against label noise during training and allows target-oriented manipulation during inference. We evaluate the proposed method on %the widely used datasets Cityscapes, KITTI-STEP, ADE20k, and COCO-stuff10k.

Yuren Cong, Wentong Liao, Bodo Rosenhahn et al.

Learning similarity between scene graphs and images aims to estimate a similarity score given a scene graph and an image. There is currently no research dedicated to this task, although it is critical for scene graph generation and downstream applications. Scene graph generation is conventionally evaluated by Recall$@K$ and mean Recall$@K$, which measure the ratio of predicted triplets that appear in the human-labeled triplet set. However, such triplet-oriented metrics fail to demonstrate the overall semantic difference between a scene graph and an image and are sensitive to annotation bias and noise. Using generated scene graphs in the downstream applications is therefore limited. To address this issue, for the first time, we propose a Scene graPh-imAge coNtrastive learning framework, SPAN, that can measure the similarity between scene graphs and images. Our novel framework consists of a graph Transformer and an image Transformer to align scene graphs and their corresponding images in the shared latent space. We introduce a novel graph serialization technique that transforms a scene graph into a sequence with structural encodings. Based on our framework, we propose R-Precision measuring image retrieval accuracy as a new evaluation metric for scene graph generation. We establish new benchmarks on the Visual Genome and Open Images datasets. Extensive experiments are conducted to verify the effectiveness of SPAN, which shows great potential as a scene graph encoder.

Frederik Schubert, Carolin Benjamins, Sebastian Döhler et al.

The goal of Unsupervised Reinforcement Learning (URL) is to find a reward-agnostic prior policy on a task domain, such that the sample-efficiency on supervised downstream tasks is improved. Although agents initialized with such a prior policy can achieve a significantly higher reward with fewer samples when finetuned on the downstream task, it is still an open question how an optimal pretrained prior policy can be achieved in practice. In this work, we present POLTER (Policy Trajectory Ensemble Regularization) - a general method to regularize the pretraining that can be applied to any URL algorithm and is especially useful on data- and knowledge-based URL algorithms. It utilizes an ensemble of policies that are discovered during pretraining and moves the policy of the URL algorithm closer to its optimal prior. Our method is based on a theoretical framework, and we analyze its practical effects on a white-box benchmark, allowing us to study POLTER with full control. In our main experiments, we evaluate POLTER on the Unsupervised Reinforcement Learning Benchmark (URLB), which consists of 12 tasks in 3 domains. We demonstrate the generality of our approach by improving the performance of a diverse set of data- and knowledge-based URL algorithms by 19% on average and up to 40% in the best case. Under a fair comparison with tuned baselines and tuned POLTER, we establish a new state-of-the-art for model-free methods on the URLB.

Yuren Cong, Mengmeng Xu, Christian Simon et al.

Text-to-video editing aims to edit the visual appearance of a source video conditional on textual prompts. A major challenge in this task is to ensure that all frames in the edited video are visually consistent. Most recent works apply advanced text-to-image diffusion models to this task by inflating 2D spatial attention in the U-Net into spatio-temporal attention. Although temporal context can be added through spatio-temporal attention, it may introduce some irrelevant information for each patch and therefore cause inconsistency in the edited video. In this paper, for the first time, we introduce optical flow into the attention module in the diffusion model's U-Net to address the inconsistency issue for text-to-video editing. Our method, FLATTEN, enforces the patches on the same flow path across different frames to attend to each other in the attention module, thus improving the visual consistency in the edited videos. Additionally, our method is training-free and can be seamlessly integrated into any diffusion-based text-to-video editing methods and improve their visual consistency. Experiment results on existing text-to-video editing benchmarks show that our proposed method achieves the new state-of-the-art performance. In particular, our method excels in maintaining the visual consistency in the edited videos.

Hendrik Hachmann, Bodo Rosenhahn

In this work, we present a marker-based multi-view spine tracking method that is specifically adjusted to the requirements for movements in sports. A maximal focus is on the accurate detection of markers and fast usage of the system. For this task, we take advantage of the prior knowledge of the arrangement of dots in perforated kinesiology tape. We detect the tape and its dots using a Mask R-CNN and a blob detector. Here, we can focus on detection only while skipping any image-based feature encoding or matching. We conduct a reasoning in 3D by a linear program and Markov random fields, in which the structure of the kinesiology tape is modeled and the shape of the spine is optimized. In comparison to state-of-the-art systems, we demonstrate that our system achieves high precision and marker density, is robust against occlusions, and capable of capturing fast movements.

Hendrik Hachmann, Bodo Rosenhahn

Deep learning-based alpha matting showed tremendous improvements in recent years, yet, feature film production studios still rely on classical chroma keying including costly post-production steps. This perceived discrepancy can be explained by some missing links necessary for production which are currently not adequately addressed in the alpha matting community, in particular foreground color estimation or color spill compensation. We propose a neural network-based temporal multi-backdrop production system that combines beneficial features from chroma keying and alpha matting. Given two consecutive frames with different background colors, our one-encoder-dual-decoder network predicts foreground colors and alpha values using a patch-based overlap-blend approach. The system is able to handle imprecise backdrops, dynamic cameras, and dynamic foregrounds and has no restrictions on foreground colors. We compare our method to state-of-the-art algorithms using benchmark datasets and a video sequence captured by a demonstrator setup. We verify that a dual backdrop input is superior to the usually applied trimap-based approach. In addition, the proposed studio set is actor friendly, and produces high-quality, temporal consistent alpha and color estimations that include a superior color spill compensation.

Vaibhav Vavilala, Florian Kluger, Seemandhar Jain et al.

Describing a scene in terms of primitives -- geometrically simple shapes that offer a parsimonious but accurate abstraction of structure -- is an established and difficult fitting problem. Different scenes require different numbers of primitives, and these primitives interact strongly. Existing methods are evaluated by comparing predicted depth, normals, and segmentation against ground truth. The state of the art method involves a learned regression procedure to predict a start point consisting of a fixed number of primitives, followed by a descent method to refine the geometry and remove redundant primitives. CSG (Constructive Solid Geometry) representations are significantly enhanced by a set-differencing operation. Our representation incorporates negative primitives, which are differenced from the positive primitives. These notably enrich the geometry that the model can encode, while complicating the fitting problem. This paper presents a method that can (a) incorporate these negative primitives and (b) choose the overall number of positive and negative primitives by ensembling. Extensive experiments on the standard NYUv2 dataset confirm that (a) this approach results in substantial improvements in depth representation and segmentation over SOTA and (b) negative primitives improve fitting accuracy. Our method is robustly applicable across datasets: in a first, we evaluate primitive prediction for LAION images.

Thomas Norrenbrock, Marco Rudolph, Bodo Rosenhahn

Explanations in Computer Vision are often desired, but most Deep Neural Networks can only provide saliency maps with questionable faithfulness. Self-Explaining Neural Networks (SENN) extract interpretable concepts with fidelity, diversity, and grounding to combine them linearly for decision-making. While they can explain what was recognized, initial realizations lack accuracy and general applicability. We propose the Quantized-Self-Explaining Neural Network Q-SENN. Q-SENN satisfies or exceeds the desiderata of SENN while being applicable to more complex datasets and maintaining most or all of the accuracy of an uninterpretable baseline model, out-performing previous work in all considered metrics. Q-SENN describes the relationship between every class and feature as either positive, negative or neutral instead of an arbitrary number of possible relations, enforcing more binary human-friendly features. Since every class is assigned just 5 interpretable features on average, Q-SENN shows convincing local and global interpretability. Additionally, we propose a feature alignment method, capable of aligning learned features with human language-based concepts without additional supervision. Thus, what is learned can be more easily verbalized. The code is published: https://github.com/ThomasNorr/Q-SENN

Tom Wehrbein, Marco Rudolph, Bodo Rosenhahn et al.

Monocular 3D human pose and shape estimation is an inherently ill-posed problem due to depth ambiguities, occlusions, and truncations. Recent probabilistic approaches learn a distribution over plausible 3D human meshes by maximizing the likelihood of the ground-truth pose given an image. We show that this objective function alone is not sufficient to best capture the full distributions. Instead, we propose to additionally supervise the learned distributions by minimizing the distance to distributions encoded in heatmaps of a 2D pose detector. Moreover, we reveal that current methods often generate incorrect hypotheses for invisible joints which is not detected by the evaluation protocols. We demonstrate that person segmentation masks can be utilized during training to significantly decrease the number of invalid samples and introduce two metrics to evaluate it. Our normalizing flow-based approach predicts plausible 3D human mesh hypotheses that are consistent with the image evidence while maintaining high diversity for ambiguous body parts. Experiments on 3DPW and EMDB show that we outperform other state-of-the-art probabilistic methods. Code is available for research purposes at https://github.com/twehrbein/humr.

Tom Wehrbein, Bodo Rosenhahn

State-of-the-art methods can recover accurate overall 3D human body motion from in-the-wild videos. However, they often fail to capture fine-grained articulations, especially in the feet, which are critical for applications such as gait analysis and animation. This limitation results from training datasets with inaccurate foot annotations and limited foot motion diversity. We address this gap with FootMR, a Foot Motion Refinement method that refines foot motion estimated by an existing human recovery model through lifting 2D foot keypoint sequences to 3D. By avoiding direct image input, FootMR circumvents inaccurate image-3D annotation pairs and can instead leverage large-scale motion capture data. To resolve ambiguities of 2D-to-3D lifting, FootMR incorporates knee and foot motion as context and predicts only residual foot motion. Generalization to extreme foot poses is further improved by representing joints in global rather than parent-relative rotations and applying extensive data augmentation. To support evaluation of foot motion reconstruction, we introduce MOOF, a 2D dataset of complex foot movements. Experiments on MOOF, MOYO, and RICH show that FootMR outperforms state-of-the-art methods, reducing ankle joint angle error on MOYO by up to 30% over the best video-based approach.

Carolin Benjamins, Theresa Eimer, Frederik Schubert et al.

While Reinforcement Learning has made great strides towards solving ever more complicated tasks, many algorithms are still brittle to even slight changes in their environment. This is a limiting factor for real-world applications of RL. Although the research community continuously aims at improving both robustness and generalization of RL algorithms, unfortunately it still lacks an open-source set of well-defined benchmark problems based on a consistent theoretical framework, which allows comparing different approaches in a fair, reliable and reproducibleway. To fill this gap, we propose CARL, a collection of well-known RL environments extended to contextual RL problems to study generalization. We show the urgent need of such benchmarks by demonstrating that even simple toy environments become challenging for commonly used approaches if different contextual instances of this task have to be considered. Furthermore, CARL allows us to provide first evidence that disentangling representation learning of the states from the policy learning with the context facilitates better generalization. By providing variations of diverse benchmarks from classic control, physical simulations, games and a real-world application of RNA design, CARL will allow the community to derive many more such insights on a solid empirical foundation.

Tom Wehrbein, Marco Rudolph, Bodo Rosenhahn et al.

3D human pose estimation from monocular images is a highly ill-posed problem due to depth ambiguities and occlusions. Nonetheless, most existing works ignore these ambiguities and only estimate a single solution. In contrast, we generate a diverse set of hypotheses that represents the full posterior distribution of feasible 3D poses. To this end, we propose a normalizing flow based method that exploits the deterministic 3D-to-2D mapping to solve the ambiguous inverse 2D-to-3D problem. Additionally, uncertain detections and occlusions are effectively modeled by incorporating uncertainty information of the 2D detector as condition. Further keys to success are a learned 3D pose prior and a generalization of the best-of-M loss. We evaluate our approach on the two benchmark datasets Human3.6M and MPI-INF-3DHP, outperforming all comparable methods in most metrics. The implementation is available on GitHub.

Yuren Cong, Wentong Liao, Hanno Ackermann et al.

Dynamic scene graph generation aims at generating a scene graph of the given video. Compared to the task of scene graph generation from images, it is more challenging because of the dynamic relationships between objects and the temporal dependencies between frames allowing for a richer semantic interpretation. In this paper, we propose Spatial-temporal Transformer (STTran), a neural network that consists of two core modules: (1) a spatial encoder that takes an input frame to extract spatial context and reason about the visual relationships within a frame, and (2) a temporal decoder which takes the output of the spatial encoder as input in order to capture the temporal dependencies between frames and infer the dynamic relationships. Furthermore, STTran is flexible to take varying lengths of videos as input without clipping, which is especially important for long videos. Our method is validated on the benchmark dataset Action Genome (AG). The experimental results demonstrate the superior performance of our method in terms of dynamic scene graphs. Moreover, a set of ablative studies is conducted and the effect of each proposed module is justified. Code available at: https://github.com/yrcong/STTran.

Hao Cheng, Wentong Liao, Xuejiao Tang et al.

To accurately predict future positions of different agents in traffic scenarios is crucial for safely deploying intelligent autonomous systems in the real-world environment. However, it remains a challenge due to the behavior of a target agent being affected by other agents dynamically and there being more than one socially possible paths the agent could take. In this paper, we propose a novel framework, named Dynamic Context Encoder Network (DCENet). In our framework, first, the spatial context between agents is explored by using self-attention architectures. Then, the two-stream encoders are trained to learn temporal context between steps by taking the respective observed trajectories and the extracted dynamic spatial context as input. The spatial-temporal context is encoded into a latent space using a Conditional Variational Auto-Encoder (CVAE) module. Finally, a set of future trajectories for each agent is predicted conditioned on the learned spatial-temporal context by sampling from the latent space, repeatedly. DCENet is evaluated on one of the most popular challenging benchmarks for trajectory forecasting Trajnet and reports a new state-of-the-art performance. It also demonstrates superior performance evaluated on the benchmark inD for mixed traffic at intersections. A series of ablation studies is conducted to validate the effectiveness of each proposed module. Our code is available at https://github.com/wtliao/DCENet.

Mathis Kruse, Marco Rudolph, Dominik Woiwode et al.

Detecting anomalies in images has become a well-explored problem in both academia and industry. State-of-the-art algorithms are able to detect defects in increasingly difficult settings and data modalities. However, most current methods are not suited to address 3D objects captured from differing poses. While solutions using Neural Radiance Fields (NeRFs) have been proposed, they suffer from excessive computation requirements, which hinder real-world usability. For this reason, we propose the novel 3D Gaussian splatting-based framework SplatPose which, given multi-view images of a 3D object, accurately estimates the pose of unseen views in a differentiable manner, and detects anomalies in them. We achieve state-of-the-art results in both training and inference speed, and detection performance, even when using less training data than competing methods. We thoroughly evaluate our framework using the recently proposed Pose-agnostic Anomaly Detection benchmark and its multi-pose anomaly detection (MAD) data set.

Melanie Schaller, Mathis Kruse, Antonio Ortega et al.

Addressing sensor drift is essential in industrial measurement systems, where precise data output is necessary for maintaining accuracy and reliability in monitoring processes, as it progressively degrades the performance of machine learning models over time. Our findings indicate that the standard cross-validation method used in existing model training overestimates performance by inadequately accounting for drift. This is primarily because typical cross-validation techniques allow data instances to appear in both training and testing sets, thereby distorting the accuracy of the predictive evaluation. As a result, these models are unable to precisely predict future drift effects, compromising their ability to generalize and adapt to evolving data conditions. This paper presents two solutions: (1) a novel sensor drift compensation learning paradigm for validating models, and (2) automated machine learning (AutoML) techniques to enhance classification performance and compensate sensor drift. By employing strategies such as data balancing, meta-learning, automated ensemble learning, hyperparameter optimization, feature selection, and boosting, our AutoML-DC (Drift Compensation) model significantly improves classification performance against sensor drift. AutoML-DC further adapts effectively to varying drift severities.

Florian Kluger, Eric Brachmann, Michael Ying Yang et al.

Humans perceive and construct the world as an arrangement of simple parametric models. In particular, we can often describe man-made environments using volumetric primitives such as cuboids or cylinders. Inferring these primitives is important for attaining high-level, abstract scene descriptions. Previous approaches for primitive-based abstraction estimate shape parameters directly and are only able to reproduce simple objects. In contrast, we propose a robust estimator for primitive fitting, which meaningfully abstracts complex real-world environments using cuboids. A RANSAC estimator guided by a neural network fits these primitives to a depth map. We condition the network on previously detected parts of the scene, parsing it one-by-one. To obtain cuboids from single RGB images, we additionally optimise a depth estimation CNN end-to-end. Naively minimising point-to-primitive distances leads to large or spurious cuboids occluding parts of the scene. We thus propose an improved occlusion-aware distance metric correctly handling opaque scenes. Furthermore, we present a neural network based cuboid solver which provides more parsimonious scene abstractions while also reducing inference time. The proposed algorithm does not require labour-intensive labels, such as cuboid annotations, for training. Results on the NYU Depth v2 dataset demonstrate that the proposed algorithm successfully abstracts cluttered real-world 3D scene layouts.

Bodo Rosenhahn, Christoph Hirche

A Normalizing Flow computes a bijective mapping from an arbitrary distribution to a predefined (e.g. normal) distribution. Such a flow can be used to address different tasks, e.g. anomaly detection, once such a mapping has been learned. In this work we introduce Normalizing Flows for Quantum architectures, describe how to model and optimize such a flow and evaluate our method on example datasets. Our proposed models show competitive performance for anomaly detection compared to classical methods, esp. those ones where there are already quantum inspired algorithms available. In the experiments we compare our performance to isolation forests (IF), the local outlier factor (LOF) or single-class SVMs.

Mariia Khan, Yue Qiu, Yuren Cong et al.

Multi-class multi-instance segmentation is the task of identifying masks for multiple object classes and multiple instances of the same class within an image. The foundational Segment Anything Model (SAM) is designed for promptable multi-class multi-instance segmentation but tends to output part or sub-part masks in the "everything" mode for various real-world applications. Whole object segmentation masks play a crucial role for indoor scene understanding, especially in robotics applications. We propose a new domain invariant Real-to-Simulation (Real-Sim) fine-tuning strategy for SAM. We use object images and ground truth data collected from Ai2Thor simulator during fine-tuning (real-to-sim). To allow our Segment Any Object Model (SAOM) to work in the "everything" mode, we propose the novel nearest neighbour assignment method, updating point embeddings for each ground-truth mask. SAOM is evaluated on our own dataset collected from Ai2Thor simulator. SAOM significantly improves on SAM, with a 28% increase in mIoU and a 25% increase in mAcc for 54 frequently-seen indoor object classes. Moreover, our Real-to-Simulation fine-tuning strategy demonstrates promising generalization performance in real environments without being trained on the real-world data (sim-to-real). The dataset and the code will be released after publication.

Timo Kaiser, Thomas Norrenbrock, Bodo Rosenhahn

The introduction of the Segment Anything Model (SAM) has paved the way for numerous semantic segmentation applications. For several tasks, quantifying the uncertainty of SAM is of particular interest. However, the ambiguous nature of the class-agnostic foundation model SAM challenges current uncertainty quantification (UQ) approaches. This paper presents a theoretically motivated uncertainty quantification model based on a Bayesian entropy formulation jointly respecting aleatoric, epistemic, and the newly introduced task uncertainty. We use this formulation to train USAM, a lightweight post-hoc UQ method. Our model traces the root of uncertainty back to under-parameterised models, insufficient prompts or image ambiguities. Our proposed deterministic USAM demonstrates superior predictive capabilities on the SA-V, MOSE, ADE20k, DAVIS, and COCO datasets, offering a computationally cheap and easy-to-use UQ alternative that can support user-prompting, enhance semi-supervised pipelines, or balance the tradeoff between accuracy and cost efficiency.

Thomas Norrenbrock, Timo Kaiser, Sovan Biswas et al.

Understanding the classifications of deep neural networks, e.g. used in safety-critical situations, is becoming increasingly important. While recent models can locally explain a single decision, to provide a faithful global explanation about an accurate model's general behavior is a more challenging open task. Towards that goal, we introduce the Quadratic Programming Enhanced Model (QPM), which learns globally interpretable class representations. QPM represents every class with a binary assignment of very few, typically 5, features, that are also assigned to other classes, ensuring easily comparable contrastive class representations. This compact binary assignment is found using discrete optimization based on predefined similarity measures and interpretability constraints. The resulting optimal assignment is used to fine-tune the diverse features, so that each of them becomes the shared general concept between the assigned classes. Extensive evaluations show that QPM delivers unprecedented global interpretability across small and large-scale datasets while setting the state of the art for the accuracy of interpretable models.

Timo Kaiser, Maximilian Schier, Bodo Rosenhahn

Cell tracking and segmentation assist biologists in extracting insights from large-scale microscopy time-lapse data. Driven by local accuracy metrics, current tracking approaches often suffer from a lack of long-term consistency and the ability to reconstruct lineage trees correctly. To address this issue, we introduce an uncertainty estimation technique for motion estimation frameworks and extend the multi-hypothesis tracking framework. Our uncertainty estimation lifts motion representations into probabilistic spatial densities using problem-specific test-time augmentations. Moreover, we introduce a novel mitosis-aware assignment problem formulation that allows multi-hypothesis trackers to model cell splits and to resolve false associations and mitosis detections based on long-term conflicts. In our framework, explicit biological knowledge is modeled in assignment costs. We evaluate our approach on nine competitive datasets and demonstrate that we outperform the current state-of-the-art on biologically inspired metrics substantially, achieving improvements by a factor of approximately 6 and uncover new insights into the behavior of motion estimation uncertainty.

Mathis Kruse, Bodo Rosenhahn

With more well-performing anomaly detection methods proposed, many of the single-view tasks have been solved to a relatively good degree. However, real-world production scenarios often involve complex industrial products, whose properties may not be fully captured by one single image. While normalizing flow based approaches already work well in single-camera scenarios, they currently do not make use of the priors in multi-view data. We aim to bridge this gap by using these flow-based models as a strong foundation and propose Multi-Flow, a novel multi-view anomaly detection method. Multi-Flow makes use of a novel multi-view architecture, whose exact likelihood estimation is enhanced by fusing information across different views. For this, we propose a new cross-view message-passing scheme, letting information flow between neighboring views. We empirically validate it on the real-world multi-view data set Real-IAD and reach a new state-of-the-art, surpassing current baselines in both image-wise and sample-wise anomaly detection tasks.

Melanie Schaller, Sergej Hloch, Akash Nag et al.

This study investigates a pulsating fluid jet as a novel precise, minimally invasive and cold technique for bone cement removal. We utilize the pulsating fluid jet device to remove bone cement from samples designed to mimic clinical conditions. The effectiveness of long nozzles was tested to enable minimally invasive procedures. Audio signal monitoring, complemented by the State Space Model (SSM) S4D-Bio, was employed to optimize the fluid jet parameters dynamically, addressing challenges like visibility obstruction from splashing. Within our experiments, we generate a comprehensive dataset correlating various process parameters and their equivalent audio signals to material erosion. The use of SSMs yields precise control over the predictive erosion process, achieving 98.93 \% accuracy. The study demonstrates on the one hand, that the pulsating fluid jet device, coupled with advanced audio monitoring techniques, is a highly effective tool for precise bone cement removal. On the other hand, this study presents the first application of SSMs in biomedical surgery technology, marking a significant advancement in the application. This research significantly advances biomedical engineering by integrating machine learning combined with pulsating fluid jet as surgical technology, offering a novel, minimally invasive, cold and adaptive approach for bone cement removal in orthopedic applications.

Christoph Reinders, Frederik Schubert, Bodo Rosenhahn

Training deep neural networks requires datasets with a large number of annotated examples. The collection and annotation of these datasets is not only extremely expensive but also faces legal and privacy problems. These factors are a significant limitation for many real-world applications. To address this, we introduce HydraMix, a novel architecture that generates new image compositions by mixing multiple different images from the same class. HydraMix learns the fusion of the content of various images guided by a segmentation-based mixing mask in feature space and is optimized via a combination of unsupervised and adversarial training. Our data augmentation scheme allows the creation of models trained from scratch on very small datasets. We conduct extensive experiments on ciFAIR-10, STL-10, and ciFAIR-100. Additionally, we introduce a novel text-image metric to assess the generality of the augmented datasets. Our results show that HydraMix outperforms existing state-of-the-art methods for image classification on small datasets.

Tom Wehrbein, Bodo Rosenhahn, Iain Matthews et al.

Recently, regression-based methods have dominated the field of 3D human pose and shape estimation. Despite their promising results, a common issue is the misalignment between predictions and image observations, often caused by minor joint rotation errors that accumulate along the kinematic chain. To address this issue, we propose to construct dense correspondences between initial human model estimates and the corresponding images that can be used to refine the initial predictions. To this end, we utilize renderings of the 3D models to predict per-pixel 2D displacements between the synthetic renderings and the RGB images. This allows us to effectively integrate and exploit appearance information of the persons. Our per-pixel displacements can be efficiently transformed to per-visible-vertex displacements and then used for 3D model refinement by minimizing a reprojection loss. To demonstrate the effectiveness of our approach, we refine the initial 3D human mesh predictions of multiple models using different refinement procedures on 3DPW and RICH. We show that our approach not only consistently leads to better image-model alignment, but also to improved 3D accuracy.

Yannik Mahlau, Frederik Schubert, Bodo Rosenhahn

The combination of self-play and planning has achieved great successes in sequential games, for instance in Chess and Go. However, adapting algorithms such as AlphaZero to simultaneous games poses a new challenge. In these games, missing information about concurrent actions of other agents is a limiting factor as they may select different Nash equilibria or do not play optimally at all. Thus, it is vital to model the behavior of the other agents when interacting with them in simultaneous games. To this end, we propose Albatross: AlphaZero for Learning Bounded-rational Agents and Temperature-based Response Optimization using Simulated Self-play. Albatross learns to play the novel equilibrium concept of a Smooth Best Response Logit Equilibrium (SBRLE), which enables cooperation and competition with agents of any playing strength. We perform an extensive evaluation of Albatross on a set of cooperative and competitive simultaneous perfect-information games. In contrast to AlphaZero, Albatross is able to exploit weak agents in the competitive game of Battlesnake. Additionally, it yields an improvement of 37.6% compared to previous state of the art in the cooperative Overcooked benchmark.

Melanie Schaller, Nick Janssen, Bodo Rosenhahn

This paper presents Naga, a deep State Space Model (SSM) encoding approach inspired by structural concepts from Vedic mathematics. The proposed method introduces a bidirectional representation for time series by jointly processing forward and time-reversed input sequences. These representations are then combined through an element-wise (Hadamard) interaction, resulting in a Vedic-inspired encoding that enhances the model's ability to capture temporal dependencies across distant time steps. We evaluate Naga on multiple long-term time series forecasting (LTSF) benchmarks, including ETTh1, ETTh2, ETTm1, ETTm2, Weather, Traffic, and ILI. The experimental results show that Naga outperforms 28 current state of the art models and demonstrates improved efficiency compared to existing deep SSM-based approaches. The findings suggest that incorporating structured, Vedic-inspired decomposition can provide an interpretable and computationally efficient alternative for long-range sequence modeling.

Thomas Norrenbrock, Timo Kaiser, Sovan Biswas et al.

Globally interpretable models are a promising approach for trustworthy AI in safety-critical domains. Alongside global explanations, detailed local explanations are a crucial complement to effectively support human experts during inference. This work proposes the Calibrated Hierarchical QPM (CHiQPM) which offers uniquely comprehensive global and local interpretability, paving the way for human-AI complementarity. CHiQPM achieves superior global interpretability by contrastively explaining the majority of classes and offers novel hierarchical explanations that are more similar to how humans reason and can be traversed to offer a built-in interpretable Conformal prediction (CP) method. Our comprehensive evaluation shows that CHiQPM achieves state-of-the-art accuracy as a point predictor, maintaining 99% accuracy of non-interpretable models. This demonstrates a substantial improvement, where interpretability is incorporated without sacrificing overall accuracy. Furthermore, its calibrated set prediction is competitively efficient to other CP methods, while providing interpretable predictions of coherent sets along its hierarchical explanation.

Bodo Rosenhahn, Tobias J. Osborne, Christoph Hirche

Translating a general quantum circuit on a specific hardware topology with a reduced set of available gates, also known as transpilation, comes with a substantial increase in the length of the equivalent circuit. Due to decoherence, the quality of the computational outcome can degrade seriously with increasing circuit length. Thus, there is major interest to reduce a quantum circuit to an equivalent circuit which is in its gate count as short as possible. One method to address efficient transpilation is based on approaches known from stochastic optimization, e.g. by using random sampling and token replacement strategies. Here, a core challenge is that these methods can suffer from sampling efficiency, causing long and energy consuming optimization time. As a remedy, we propose in this work 2D neural guided sampling. Thus, given a 2D representation of a quantum circuit, a neural network predicts groups of gates in the quantum circuit, which are likely reducible. Thus, it leads to a sampling prior which can heavily reduce the compute time for quantum circuit reduction. In several experiments, we demonstrate that our method is superior to results obtained from different qiskit or BQSKit optimization levels.

Nick Janßen, Melanie Schaller, Bodo Rosenhahn

Understanding the robustness of deep learning models for multivariate long-term time series forecasting (M-LTSF) remains challenging, as evaluations typically rely on real-world datasets with unknown noise properties. We propose a simulation-based evaluation framework that generates parameterizable synthetic datasets, where each dataset instance corresponds to a different configuration of signal components, noise types, signal-to-noise ratios, and frequency characteristics. These configurable components aim to model real-world multivariate time series data without the ambiguity of unknown noise. This framework enables fine-grained, systematic evaluation of M-LTSF models under controlled and diverse scenarios. We benchmark four representative architectures S-Mamba (state-space), iTransformer (transformer-based), R-Linear (linear), and Autoformer (decomposition-based). Our analysis reveals that all models degrade severely when lookback windows cannot capture complete periods of seasonal patters in the data. S-Mamba and Autoformer perform best on sawtooth patterns, while R-Linear and iTransformer favor sinusoidal signals. White and Brownian noise universally degrade performance with lower signal-to-noise ratio while S-Mamba shows specific trend-noise and iTransformer shows seasonal-noise vulnerability. Further spectral analysis shows that S-Mamba and iTransformer achieve superior frequency reconstruction. This controlled approach, based on our synthetic and principle-driven testbed, offers deeper insights into model-specific strengths and limitations through the aggregation of MSE scores and provides concrete guidance for model selection based on signal characteristics and noise conditions.

Patrick Glandorf, Bodo Rosenhahn

Vision Transformer have set new benchmarks in several tasks, but these models come with the lack of high computational costs which makes them impractical for resource limited hardware. Network pruning reduces the computational complexity by removing less important operations while maintaining performance. However, pruning a model on an unseen data domain, leads to a misevaluation of weight significance, resulting in suboptimal resource assignment. In this work, we find that task-sensitive layers initially fail to improve the feature representation on downstream tasks, leading to performance loss for early pruning decisions. To address this problem, we introduce Pruning by Block Benefit (P3B), a pruning method that utilizes the relative contribution on block level to globally assign parameter resources. P3B identifies low-impact components to reduce parameter allocation while preserving critical ones. Classical pruning mask optimization struggles to reactivate zero-mask-elements. In contrast, P3B sets a layerwise keep ratio based on global performance metrics, ensuring the reactivation of late-converging blocks. We show in extensive experiments that P3B is a state of the art pruning method with most noticeable gains in transfer learning tasks. Notably, P3B is able to conserve high performance, even in high sparsity regimes of 70% parameter reduction while only losing 0.64% in accuracy.

Florian Kluger, Bodo Rosenhahn

We present a real-time method for robust estimation of multiple instances of geometric models from noisy data. Geometric models such as vanishing points, planar homographies or fundamental matrices are essential for 3D scene analysis. Previous approaches discover distinct model instances in an iterative manner, thus limiting their potential for speedup via parallel computation. In contrast, our method detects all model instances independently and in parallel. A neural network segments the input data into clusters representing potential model instances by predicting multiple sets of sample and inlier weights. Using the predicted weights, we determine the model parameters for each potential instance separately in a RANSAC-like fashion. We train the neural network via task-specific loss functions, i.e. we do not require a ground-truth segmentation of the input data. As suitable training data for homography and fundamental matrix fitting is scarce, we additionally present two new synthetic datasets. We demonstrate state-of-the-art performance on these as well as multiple established datasets, with inference times as small as five milliseconds per image.

Christoph Reinders, Frederik Schubert, Bodo Rosenhahn

Deep convolutional neural networks require large amounts of labeled data samples. For many real-world applications, this is a major limitation which is commonly treated by augmentation methods. In this work, we address the problem of learning deep neural networks on small datasets. Our proposed architecture called ChimeraMix learns a data augmentation by generating compositions of instances. The generative model encodes images in pairs, combines the features guided by a mask, and creates new samples. For evaluation, all methods are trained from scratch without any additional data. Several experiments on benchmark datasets, e.g. ciFAIR-10, STL-10, and ciFAIR-100, demonstrate the superior performance of ChimeraMix compared to current state-of-the-art methods for classification on small datasets.

Carolin Benjamins, Theresa Eimer, Frederik Schubert et al.

While Reinforcement Learning ( RL) has made great strides towards solving increasingly complicated problems, many algorithms are still brittle to even slight environmental changes. Contextual Reinforcement Learning (cRL) provides a framework to model such changes in a principled manner, thereby enabling flexible, precise and interpretable task specification and generation. Our goal is to show how the framework of cRL contributes to improving zero-shot generalization in RL through meaningful benchmarks and structured reasoning about generalization tasks. We confirm the insight that optimal behavior in cRL requires context information, as in other related areas of partial observability. To empirically validate this in the cRL framework, we provide various context-extended versions of common RL environments. They are part of the first benchmark library, CARL, designed for generalization based on cRL extensions of popular benchmarks, which we propose as a testbed to further study general agents. We show that in the contextual setting, even simple RL environments become challenging - and that naive solutions are not enough to generalize across complex context spaces.

Yuren Cong, Michael Ying Yang, Bodo Rosenhahn

Different objects in the same scene are more or less related to each other, but only a limited number of these relationships are noteworthy. Inspired by DETR, which excels in object detection, we view scene graph generation as a set prediction problem and propose an end-to-end scene graph generation model RelTR which has an encoder-decoder architecture. The encoder reasons about the visual feature context while the decoder infers a fixed-size set of triplets subject-predicate-object using different types of attention mechanisms with coupled subject and object queries. We design a set prediction loss performing the matching between the ground truth and predicted triplets for the end-to-end training. In contrast to most existing scene graph generation methods, RelTR is a one-stage method that predicts a set of relationships directly only using visual appearance without combining entities and labeling all possible predicates. Extensive experiments on the Visual Genome and Open Images V6 datasets demonstrate the superior performance and fast inference of our model.

Duy M. H. Nguyen, Roberto Henschel, Bodo Rosenhahn et al.

Multi-Camera Multi-Object Tracking is currently drawing attention in the computer vision field due to its superior performance in real-world applications such as video surveillance in crowded scenes or in wide spaces. In this work, we propose a mathematically elegant multi-camera multiple object tracking approach based on a spatial-temporal lifted multicut formulation. Our model utilizes state-of-the-art tracklets produced by single-camera trackers as proposals. As these tracklets may contain ID-Switch errors, we refine them through a novel pre-clustering obtained from 3D geometry projections. As a result, we derive a better tracking graph without ID switches and more precise affinity costs for the data association phase. Tracklets are then matched to multi-camera trajectories by solving a global lifted multicut formulation that incorporates short and long-range temporal interactions on tracklets located in the same camera as well as inter-camera ones. Experimental results on the WildTrack dataset yield near-perfect performance, outperforming state-of-the-art trackers on Campus while being on par on the PETS-09 dataset.