Stefan Roth

Gopika Sudhakaran, Devendra Singh Dhami, Kristian Kersting et al.

Recent years have seen a growing interest in Scene Graph Generation (SGG), a comprehensive visual scene understanding task that aims to predict entity relationships using a relation encoder-decoder pipeline stacked on top of an object encoder-decoder backbone. Unfortunately, current SGG methods suffer from an information loss regarding the entities local-level cues during the relation encoding process. To mitigate this, we introduce the Vision rElation TransfOrmer (VETO), consisting of a novel local-level entity relation encoder. We further observe that many existing SGG methods claim to be unbiased, but are still biased towards either head or tail classes. To overcome this bias, we introduce a Mutually Exclusive ExperT (MEET) learning strategy that captures important relation features without bias towards head or tail classes. Experimental results on the VG and GQA datasets demonstrate that VETO + MEET boosts the predictive performance by up to 47 percentage over the state of the art while being 10 times smaller.

Moritz Nottebaum, Stefan Roth, Simone Schaub-Meyer

Most deep learning methods for video frame interpolation consist of three main components: feature extraction, motion estimation, and image synthesis. Existing approaches are mainly distinguishable in terms of how these modules are designed. However, when interpolating high-resolution images, e.g. at 4K, the design choices for achieving high accuracy within reasonable memory requirements are limited. The feature extraction layers help to compress the input and extract relevant information for the latter stages, such as motion estimation. However, these layers are often costly in parameters, computation time, and memory. We show how ideas from dimensionality reduction combined with a lightweight optimization can be used to compress the input representation while keeping the extracted information suitable for frame interpolation. Further, we require neither a pretrained flow network nor a synthesis network, additionally reducing the number of trainable parameters and required memory. When evaluating on three 4K benchmarks, we achieve state-of-the-art image quality among the methods without pretrained flow while having the lowest network complexity and memory requirements overall.

Franz Klein, Shweta Mahajan, Stefan Roth

Stylized image captioning as presented in prior work aims to generate captions that reflect characteristics beyond a factual description of the scene composition, such as sentiments. Such prior work relies on given sentiment identifiers, which are used to express a certain global style in the caption, e.g. positive or negative, however without taking into account the stylistic content of the visual scene. To address this shortcoming, we first analyze the limitations of current stylized captioning datasets and propose COCO attribute-based augmentations to obtain varied stylized captions from COCO annotations. Furthermore, we encode the stylized information in the latent space of a Variational Autoencoder; specifically, we leverage extracted image attributes to explicitly structure its sequential latent space according to different localized style characteristics. Our experiments on the Senticap and COCO datasets show the ability of our approach to generate accurate captions with diversity in styles that are grounded in the image.

Krishnakant Singh, Simone Schaub-Meyer, Stefan Roth

The high-quality images yielded by generative adversarial networks (GANs) have motivated investigations into their application for image editing. However, GANs are often limited in the control they provide for performing specific edits. One of the principal challenges is the entangled latent space of GANs, which is not directly suitable for performing independent and detailed edits. Recent editing methods allow for either controlled style edits or controlled semantic edits. In addition, methods that use semantic masks to edit images have difficulty preserving the identity and are unable to perform controlled style edits. We propose a method to disentangle a GAN$\text{'}$s latent space into semantic and style spaces, enabling controlled semantic and style edits for face images independently within the same framework. To achieve this, we design an encoder-decoder based network architecture ($S^2$-Flow), which incorporates two proposed inductive biases. We show the suitability of $S^2$-Flow quantitatively and qualitatively by performing various semantic and style edits.

Krishnakant Singh, Simone Schaub-Meyer, Stefan Roth

Object-centric learning aims to decompose an input image into a set of meaningful object files (slots). These latent object representations enable a variety of downstream tasks. Yet, object-centric learning struggles on real-world datasets, which contain multiple objects of complex textures and shapes in natural everyday scenes. To address this, we introduce Guided Latent Slot Diffusion (GLASS), a novel slot-attention model that learns in the space of generated images and uses semantic and instance guidance modules to learn better slot embeddings for various downstream tasks. Our experiments show that GLASS surpasses state-of-the-art slot-attention methods by a wide margin on tasks such as (zero-shot) object discovery and conditional image generation for real-world scenes. Moreover, GLASS enables the first application of slot attention to the compositional generation of complex, realistic scenes.

Tobias Lingenberg, Markus Reuter, Gopika Sudhakaran et al.

Simple data augmentation techniques, such as rotations and flips, are widely used to enhance the generalization power of computer vision models. However, these techniques often fail to modify high-level semantic attributes of a class. To address this limitation, researchers have explored generative augmentation methods like the recently proposed DA-Fusion. Despite some progress, the variations are still largely limited to textural changes, thus falling short on aspects like varied viewpoints, environment, weather conditions, or even class-level semantic attributes (eg, variations in a dog's breed). To overcome this challenge, we propose DIAGen, building upon DA-Fusion. First, we apply Gaussian noise to the embeddings of an object learned with Textual Inversion to diversify generations using a pre-trained diffusion model's knowledge. Second, we exploit the general knowledge of a text-to-text generative model to guide the image generation of the diffusion model with varied class-specific prompts. Finally, we introduce a weighting mechanism to mitigate the impact of poorly generated samples. Experimental results across various datasets show that DIAGen not only enhances semantic diversity but also improves the performance of subsequent classifiers. The advantages of DIAGen over standard augmentations and the DA-Fusion baseline are particularly pronounced with out-of-distribution samples.

Oliver Hahn, Nikita Araslanov, Simone Schaub-Meyer et al.

Unsupervised semantic segmentation aims to automatically partition images into semantically meaningful regions by identifying global semantic categories within an image corpus without any form of annotation. Building upon recent advances in self-supervised representation learning, we focus on how to leverage these large pre-trained models for the downstream task of unsupervised segmentation. We present PriMaPs - Principal Mask Proposals - decomposing images into semantically meaningful masks based on their feature representation. This allows us to realize unsupervised semantic segmentation by fitting class prototypes to PriMaPs with a stochastic expectation-maximization algorithm, PriMaPs-EM. Despite its conceptual simplicity, PriMaPs-EM leads to competitive results across various pre-trained backbone models, including DINO and DINOv2, and across different datasets, such as Cityscapes, COCO-Stuff, and Potsdam-3. Importantly, PriMaPs-EM is able to boost results when applied orthogonally to current state-of-the-art unsupervised semantic segmentation pipelines. Code is available at https://github.com/visinf/primaps.

Sascha Rosbach, Stefan M. Leupold, Simon Großjohann et al.

Automated vehicles operating in urban environments have to reliably interact with other traffic participants. Planning algorithms often utilize separate prediction modules forecasting probabilistic, multi-modal, and interactive behaviors of objects. Designing prediction and planning as two separate modules introduces significant challenges, particularly due to the interdependence of these modules. This work proposes a deep learning methodology to combine prediction and planning. A conditional GAN with the U-Net architecture is trained to predict two high-resolution image sequences. The sequences represent explicit motion predictions, mainly used to train context understanding, and pixel state values suitable for planning encoding kinematic reachability, object dynamics, safety, and driving comfort. The model can be trained offline on target images rendered by a sampling-based model-predictive planner, leveraging real-world driving data. Our results demonstrate intuitive behavior in complex situations, such as lane changes amidst conflicting objectives.

Robin Hesse, Jonas Fischer, Simone Schaub-Meyer et al.

Mechanistic interpretability is concerned with analyzing individual components in a (convolutional) neural network (CNN) and how they form larger circuits representing decision mechanisms. These investigations are challenging since CNNs frequently learn polysemantic channels that encode distinct concepts, making them hard to interpret. To address this, we propose an algorithm to disentangle a specific kind of polysemantic channel into multiple channels, each responding to a single concept. Our approach restructures weights in a CNN, utilizing that different concepts within the same channel exhibit distinct activation patterns in the previous layer. By disentangling these polysemantic features, we enhance the interpretability of CNNs, ultimately improving explanatory techniques such as feature visualizations.

Simon Kiefhaber, Stefan Roth, Simone Schaub-Meyer

Cost volumes are used in every modern optical flow estimator, but due to their computational and space complexity, they are often a limiting factor regarding both processing speed and the resolution of input frames. Motivated by our empirical observation that cost volumes lose their importance once all other network parts of, e.g., a RAFT-based pipeline have been sufficiently trained, we introduce a training strategy that allows removing the cost volume from optical flow estimators throughout training. This leads to significantly improved inference speed and reduced memory requirements. Using our training strategy, we create three different models covering different compute budgets. Our most accurate model reaches state-of-the-art accuracy while being $1.2\times$ faster and having a $6\times$ lower memory footprint than comparable models; our fastest model is capable of processing Full HD frames at $20\,\mathrm{FPS}$ using only $500\,\mathrm{MB}$ of GPU memory.

Barış Zöngür, Robin Hesse, Stefan Roth

To ensure the reliability of deep models in real-world applications, out-of-distribution (OOD) detection methods aim to distinguish samples close to the training distribution (in-distribution, ID) from those farther away (OOD). In this work, we propose a novel OOD detection method that utilizes singular value decomposition of the weight matrix of the classification head to decompose the model's activations into decisive and insignificant components, which contribute maximally, respectively minimally, to the final classifier output. We find that the subspace of insignificant components more effectively distinguishes ID from OOD data than raw activations in regimes of large distribution shifts (Far-OOD). This occurs because the classification objective leaves the insignificant subspace largely unaffected, yielding features that are ''untainted'' by the target classification task. Conversely, in regimes of smaller distribution shifts (Near-OOD), we find that activation shaping methods profit from only considering the decisive subspace, as the insignificant component can cause interference in the activation space. By combining two findings into a single approach, termed ActSub, we achieve state-of-the-art results in various standard OOD benchmarks.

Robin Hesse, Simone Schaub-Meyer, Stefan Roth

Many convolutional neural networks (CNNs) rely on progressive downsampling of their feature maps to increase the network's receptive field and decrease computational cost. However, this comes at the price of losing granularity in the feature maps, limiting the ability to correctly understand images or recover fine detail in dense prediction tasks. To address this, common practice is to replace the last few downsampling operations in a CNN with dilated convolutions, allowing to retain the feature map resolution without reducing the receptive field, albeit increasing the computational cost. This allows to trade off predictive performance against cost, depending on the output feature resolution. By either regularly downsampling or not downsampling the entire feature map, existing work implicitly treats all regions of the input image and subsequent feature maps as equally important, which generally does not hold. We propose an adaptive downsampling scheme that generalizes the above idea by allowing to process informative regions at a higher resolution than less informative ones. In a variety of experiments, we demonstrate the versatility of our adaptive downsampling strategy and empirically show that it improves the cost-accuracy trade-off of various established CNNs.

Robin Hesse, Simone Schaub-Meyer, Stefan Roth

Mitigating the dependence on spurious correlations present in the training dataset is a quickly emerging and important topic of deep learning. Recent approaches include priors on the feature attribution of a deep neural network (DNN) into the training process to reduce the dependence on unwanted features. However, until now one needed to trade off high-quality attributions, satisfying desirable axioms, against the time required to compute them. This in turn either led to long training times or ineffective attribution priors. In this work, we break this trade-off by considering a special class of efficiently axiomatically attributable DNNs for which an axiomatic feature attribution can be computed with only a single forward/backward pass. We formally prove that nonnegatively homogeneous DNNs, here termed $\mathcal{X}$-DNNs, are efficiently axiomatically attributable and show that they can be effortlessly constructed from a wide range of regular DNNs by simply removing the bias term of each layer. Various experiments demonstrate the advantages of $\mathcal{X}$-DNNs, beating state-of-the-art generic attribution methods on regular DNNs for training with attribution priors.

Nikita Araslanov, Simone Schaub-Meyer, Stefan Roth

We present a novel approach to unsupervised learning for video object segmentation (VOS). Unlike previous work, our formulation allows to learn dense feature representations directly in a fully convolutional regime. We rely on uniform grid sampling to extract a set of anchors and train our model to disambiguate between them on both inter- and intra-video levels. However, a naive scheme to train such a model results in a degenerate solution. We propose to prevent this with a simple regularisation scheme, accommodating the equivariance property of the segmentation task to similarity transformations. Our training objective admits efficient implementation and exhibits fast training convergence. On established VOS benchmarks, our approach exceeds the segmentation accuracy of previous work despite using significantly less training data and compute power.

Shweta Mahajan, Stefan Roth

Autoregressive models are a class of exact inference approaches with highly flexible functional forms, yielding state-of-the-art density estimates for natural images. Yet, the sequential ordering on the dimensions makes these models computationally expensive and limits their applicability to low-resolution imagery. In this work, we propose Pixel-Pyramids, a block-autoregressive approach employing a lossless pyramid decomposition with scale-specific representations to encode the joint distribution of image pixels. Crucially, it affords a sparser dependency structure compared to fully autoregressive approaches. Our PixelPyramids yield state-of-the-art results for density estimation on various image datasets, especially for high-resolution data. For CelebA-HQ 1024 x 1024, we observe that the density estimates (in terms of bits/dim) are improved to ~44% of the baseline despite sampling speeds superior even to easily parallelizable flow-based models.

Jonas Pfeiffer, Gregor Geigle, Aishwarya Kamath et al.

Recent advances in multimodal vision and language modeling have predominantly focused on the English language, mostly due to the lack of multilingual multimodal datasets to steer modeling efforts. In this work, we address this gap and provide xGQA, a new multilingual evaluation benchmark for the visual question answering task. We extend the established English GQA dataset to 7 typologically diverse languages, enabling us to detect and explore crucial challenges in cross-lingual visual question answering. We further propose new adapter-based approaches to adapt multimodal transformer-based models to become multilingual, and -- vice versa -- multilingual models to become multimodal. Our proposed methods outperform current state-of-the-art multilingual multimodal models (e.g., M3P) in zero-shot cross-lingual settings, but the accuracy remains low across the board; a performance drop of around 38 accuracy points in target languages showcases the difficulty of zero-shot cross-lingual transfer for this task. Our results suggest that simple cross-lingual transfer of multimodal models yields latent multilingual multimodal misalignment, calling for more sophisticated methods for vision and multilingual language modeling.

Jan-Martin O. Steitz, Jonas Pfeiffer, Iryna Gurevych et al.

Reasoning over multiple modalities, e.g. in Visual Question Answering (VQA), requires an alignment of semantic concepts across domains. Despite the widespread success of end-to-end learning, today's multimodal pipelines by and large leverage pre-extracted, fixed features from object detectors, typically Faster R-CNN, as representations of the visual world. The obvious downside is that the visual representation is not specifically tuned to the multimodal task at hand. At the same time, while transformer-based object detectors have gained popularity, they have not been employed in today's multimodal pipelines. We address both shortcomings with TxT, a transformer-based crossmodal pipeline that enables fine-tuning both language and visual components on the downstream task in a fully end-to-end manner. We overcome existing limitations of transformer-based detectors for multimodal reasoning regarding the integration of global context and their scalability. Our transformer-based multimodal model achieves considerable gains from end-to-end learning for multimodal question answering.

Junhwa Hur, Stefan Roth

Estimating 3D scene flow from a sequence of monocular images has been gaining increased attention due to the simple, economical capture setup. Owing to the severe ill-posedness of the problem, the accuracy of current methods has been limited, especially that of efficient, real-time approaches. In this paper, we introduce a multi-frame monocular scene flow network based on self-supervised learning, improving the accuracy over previous networks while retaining real-time efficiency. Based on an advanced two-frame baseline with a split-decoder design, we propose (i) a multi-frame model using a triple frame input and convolutional LSTM connections, (ii) an occlusion-aware census loss for better accuracy, and (iii) a gradient detaching strategy to improve training stability. On the KITTI dataset, we observe state-of-the-art accuracy among monocular scene flow methods based on self-supervised learning.

Jiangxin Dong, Stefan Roth, Bernt Schiele

We present a simple and effective approach for non-blind image deblurring, combining classical techniques and deep learning. In contrast to existing methods that deblur the image directly in the standard image space, we propose to perform an explicit deconvolution process in a feature space by integrating a classical Wiener deconvolution framework with learned deep features. A multi-scale cascaded feature refinement module then predicts the deblurred image from the deconvolved deep features, progressively recovering detail and small-scale structures. The proposed model is trained in an end-to-end manner and evaluated on scenarios with simulated Gaussian noise, saturated pixels, or JPEG compression artifacts as well as real-world images. Moreover, we present detailed analyses of the benefit of the feature-based Wiener deconvolution and of the multi-scale cascaded feature refinement as well as the robustness of the proposed approach. Our extensive experimental results show that the proposed deep Wiener deconvolution network facilitates deblurred results with visibly fewer artifacts and quantitatively outperforms state-of-the-art non-blind image deblurring methods by a wide margin.

Shweta Mahajan, Stefan Roth

Diverse image captioning models aim to learn one-to-many mappings that are innate to cross-domain datasets, such as of images and texts. Current methods for this task are based on generative latent variable models, e.g. VAEs with structured latent spaces. Yet, the amount of multimodality captured by prior work is limited to that of the paired training data -- the true diversity of the underlying generative process is not fully captured. To address this limitation, we leverage the contextual descriptions in the dataset that explain similar contexts in different visual scenes. To this end, we introduce a novel factorization of the latent space, termed context-object split, to model diversity in contextual descriptions across images and texts within the dataset. Our framework not only enables diverse captioning through context-based pseudo supervision, but extends this to images with novel objects and without paired captions in the training data. We evaluate our COS-CVAE approach on the standard COCO dataset and on the held-out COCO dataset consisting of images with novel objects, showing significant gains in accuracy and diversity.

Patrick Dendorfer, Aljoša Ošep, Anton Milan et al.

Standardized benchmarks have been crucial in pushing the performance of computer vision algorithms, especially since the advent of deep learning. Although leaderboards should not be over-claimed, they often provide the most objective measure of performance and are therefore important guides for research. We present MOTChallenge, a benchmark for single-camera Multiple Object Tracking (MOT) launched in late 2014, to collect existing and new data, and create a framework for the standardized evaluation of multiple object tracking methods. The benchmark is focused on multiple people tracking, since pedestrians are by far the most studied object in the tracking community, with applications ranging from robot navigation to self-driving cars. This paper collects the first three releases of the benchmark: (i) MOT15, along with numerous state-of-the-art results that were submitted in the last years, (ii) MOT16, which contains new challenging videos, and (iii) MOT17, that extends MOT16 sequences with more precise labels and evaluates tracking performance on three different object detectors. The second and third release not only offers a significant increase in the number of labeled boxes but also provide labels for multiple object classes beside pedestrians, as well as the level of visibility for every single object of interest. We finally provide a categorization of state-of-the-art trackers and a broad error analysis. This will help newcomers understand the related work and research trends in the MOT community, and hopefully shed some light on potential future research directions.

Sascha Rosbach, Xing Li, Simon Großjohann et al.

General-purpose trajectory planning algorithms for automated driving utilize complex reward functions to perform a combined optimization of strategic, behavioral, and kinematic features. The specification and tuning of a single reward function is a tedious task and does not generalize over a large set of traffic situations. Deep learning approaches based on path integral inverse reinforcement learning have been successfully applied to predict local situation-dependent reward functions using features of a set of sampled driving policies. Sample-based trajectory planning algorithms are able to approximate a spatio-temporal subspace of feasible driving policies that can be used to encode the context of a situation. However, the interaction with dynamic objects requires an extended planning horizon, which depends on sequential context modeling. In this work, we are concerned with the sequential reward prediction over an extended time horizon. We present a neural network architecture that uses a policy attention mechanism to generate a low-dimensional context vector by concentrating on trajectories with a human-like driving style. Apart from this, we propose a temporal attention mechanism to identify context switches and allow for stable adaptation of rewards. We evaluate our results on complex simulated driving situations, including other moving vehicles. Our evaluation shows that our policy attention mechanism learns to focus on collision-free policies in the configuration space. Furthermore, the temporal attention mechanism learns persistent interaction with other vehicles over an extended planning horizon.

Wentong Liao, Xiang Chen, Jingfeng Yang et al.

State-of-the-art object detection approaches such as Fast/Faster R-CNN, SSD, or YOLO have difficulties detecting dense, small targets with arbitrary orientation in large aerial images. The main reason is that using interpolation to align RoI features can result in a lack of accuracy or even loss of location information. We present the Local-aware Region Convolutional Neural Network (LR-CNN), a novel two-stage approach for vehicle detection in aerial imagery. We enhance translation invariance to detect dense vehicles and address the boundary quantization issue amongst dense vehicles by aggregating the high-precision RoIs' features. Moreover, we resample high-level semantic pooled features, making them regain location information from the features of a shallower convolutional block. This strengthens the local feature invariance for the resampled features and enables detecting vehicles in an arbitrary orientation. The local feature invariance enhances the learning ability of the focal loss function, and the focal loss further helps to focus on the hard examples. Taken together, our method better addresses the challenges of aerial imagery. We evaluate our approach on several challenging datasets (VEDAI, DOTA), demonstrating a significant improvement over state-of-the-art methods. We demonstrate the good generalization ability of our approach on the DLR 3K dataset.

Nikita Araslanov, Stefan Roth

Recent years have seen a rapid growth in new approaches improving the accuracy of semantic segmentation in a weakly supervised setting, i.e. with only image-level labels available for training. However, this has come at the cost of increased model complexity and sophisticated multi-stage training procedures. This is in contrast to earlier work that used only a single stage $-$ training one segmentation network on image labels $-$ which was abandoned due to inferior segmentation accuracy. In this work, we first define three desirable properties of a weakly supervised method: local consistency, semantic fidelity, and completeness. Using these properties as guidelines, we then develop a segmentation-based network model and a self-supervised training scheme to train for semantic masks from image-level annotations in a single stage. We show that despite its simplicity, our method achieves results that are competitive with significantly more complex pipelines, substantially outperforming earlier single-stage methods.

Junhwa Hur, Stefan Roth

Scene flow estimation has been receiving increasing attention for 3D environment perception. Monocular scene flow estimation -- obtaining 3D structure and 3D motion from two temporally consecutive images -- is a highly ill-posed problem, and practical solutions are lacking to date. We propose a novel monocular scene flow method that yields competitive accuracy and real-time performance. By taking an inverse problem view, we design a single convolutional neural network (CNN) that successfully estimates depth and 3D motion simultaneously from a classical optical flow cost volume. We adopt self-supervised learning with 3D loss functions and occlusion reasoning to leverage unlabeled data. We validate our design choices, including the proxy loss and augmentation setup. Our model achieves state-of-the-art accuracy among unsupervised/self-supervised learning approaches to monocular scene flow, and yields competitive results for the optical flow and monocular depth estimation sub-tasks. Semi-supervised fine-tuning further improves the accuracy and yields promising results in real-time.

Shweta Mahajan, Apratim Bhattacharyya, Mario Fritz et al.

Flow-based generative models are an important class of exact inference models that admit efficient inference and sampling for image synthesis. Owing to the efficiency constraints on the design of the flow layers, e.g. split coupling flow layers in which approximately half the pixels do not undergo further transformations, they have limited expressiveness for modeling long-range data dependencies compared to autoregressive models that rely on conditional pixel-wise generation. In this work, we improve the representational power of flow-based models by introducing channel-wise dependencies in their latent space through multi-scale autoregressive priors (mAR). Our mAR prior for models with split coupling flow layers (mAR-SCF) can better capture dependencies in complex multimodal data. The resulting model achieves state-of-the-art density estimation results on MNIST, CIFAR-10, and ImageNet. Furthermore, we show that mAR-SCF allows for improved image generation quality, with gains in FID and Inception scores compared to state-of-the-art flow-based models.

Junhwa Hur, Stefan Roth

Akin to many subareas of computer vision, the recent advances in deep learning have also significantly influenced the literature on optical flow. Previously, the literature had been dominated by classical energy-based models, which formulate optical flow estimation as an energy minimization problem. However, as the practical benefits of Convolutional Neural Networks (CNNs) over conventional methods have become apparent in numerous areas of computer vision and beyond, they have also seen increased adoption in the context of motion estimation to the point where the current state of the art in terms of accuracy is set by CNN approaches. We first review this transition as well as the developments from early work to the current state of CNNs for optical flow estimation. Alongside, we discuss some of their technical details and compare them to recapitulate which technical contribution led to the most significant accuracy improvements. Then we provide an overview of the various optical flow approaches introduced in the deep learning age, including those based on alternative learning paradigms (e.g., unsupervised and semi-supervised methods) as well as the extension to the multi-frame case, which is able to yield further accuracy improvements.

Anne S. Wannenwetsch, Stefan Roth

Encoder-decoder networks have found widespread use in various dense prediction tasks. However, the strong reduction of spatial resolution in the encoder leads to a loss of location information as well as boundary artifacts. To address this, image-adaptive post-processing methods have shown beneficial by leveraging the high-resolution input image(s) as guidance data. We extend such approaches by considering an important orthogonal source of information: the network's confidence in its own predictions. We introduce probabilistic pixel-adaptive convolutions (PPACs), which not only depend on image guidance data for filtering, but also respect the reliability of per-pixel predictions. As such, PPACs allow for image-adaptive smoothing and simultaneously propagating pixels of high confidence into less reliable regions, while respecting object boundaries. We demonstrate their utility in refinement networks for optical flow and semantic segmentation, where PPACs lead to a clear reduction in boundary artifacts. Moreover, our proposed refinement step is able to substantially improve the accuracy on various widely used benchmarks.

Patrick Dendorfer, Hamid Rezatofighi, Anton Milan et al.

Standardized benchmarks are crucial for the majority of computer vision applications. Although leaderboards and ranking tables should not be over-claimed, benchmarks often provide the most objective measure of performance and are therefore important guides for research. The benchmark for Multiple Object Tracking, MOTChallenge, was launched with the goal to establish a standardized evaluation of multiple object tracking methods. The challenge focuses on multiple people tracking, since pedestrians are well studied in the tracking community, and precise tracking and detection has high practical relevance. Since the first release, MOT15, MOT16, and MOT17 have tremendously contributed to the community by introducing a clean dataset and precise framework to benchmark multi-object trackers. In this paper, we present our MOT20benchmark, consisting of 8 new sequences depicting very crowded challenging scenes. The benchmark was presented first at the 4thBMTT MOT Challenge Workshop at the Computer Vision and Pattern Recognition Conference (CVPR) 2019, and gives to chance to evaluate state-of-the-art methods for multiple object tracking when handling extremely crowded scenarios.

Shweta Mahajan, Iryna Gurevych, Stefan Roth

Learned joint representations of images and text form the backbone of several important cross-domain tasks such as image captioning. Prior work mostly maps both domains into a common latent representation in a purely supervised fashion. This is rather restrictive, however, as the two domains follow distinct generative processes. Therefore, we propose a novel semi-supervised framework, which models shared information between domains and domain-specific information separately. The information shared between the domains is aligned with an invertible neural network. Our model integrates normalizing flow-based priors for the domain-specific information, which allows us to learn diverse many-to-many mappings between the two domains. We demonstrate the effectiveness of our model on diverse tasks, including image captioning and text-to-image synthesis.

Vladyslav Yushchenko, Nikita Araslanov, Stefan Roth

We identify two pathological cases of temporal inconsistencies in video generation: video freezing and video looping. To better quantify the temporal diversity, we propose a class of complementary metrics that are effective, easy to implement, data agnostic, and interpretable. Further, we observe that current state-of-the-art models are trained on video samples of fixed length thereby inhibiting long-term modeling. To address this, we reformulate the problem of video generation as a Markov Decision Process (MDP). The underlying idea is to represent motion as a stochastic process with an infinite forecast horizon to overcome the fixed length limitation and to mitigate the presence of temporal artifacts. We show that our formulation is easy to integrate into the state-of-the-art MoCoGAN framework. Our experiments on the Human Actions and UCF-101 datasets demonstrate that our MDP-based model is more memory efficient and improves the video quality both in terms of the new and established metrics.

Jochen Gast, Stefan Roth

Video deblurring for hand-held cameras is a challenging task, since the underlying blur is caused by both camera shake and object motion. State-of-the-art deep networks exploit temporal information from neighboring frames, either by means of spatio-temporal transformers or by recurrent architectures. In contrast to these involved models, we found that a simple baseline CNN can perform astonishingly well when particular care is taken w.r.t. the details of model and training procedure. To that end, we conduct a comprehensive study regarding these crucial details, uncovering extreme differences in quantitative and qualitative performance. Exploiting these details allows us to boost the architecture and training procedure of a simple baseline CNN by a staggering 3.15dB, such that it becomes highly competitive w.r.t. cutting-edge networks. This raises the question whether the reported accuracy difference between models is always due to technical contributions or also subject to such orthogonal, but crucial details.

Shweta Mahajan, Teresa Botschen, Iryna Gurevych et al.

One of the key challenges in learning joint embeddings of multiple modalities, e.g. of images and text, is to ensure coherent cross-modal semantics that generalize across datasets. We propose to address this through joint Gaussian regularization of the latent representations. Building on Wasserstein autoencoders (WAEs) to encode the input in each domain, we enforce the latent embeddings to be similar to a Gaussian prior that is shared across the two domains, ensuring compatible continuity of the encoded semantic representations of images and texts. Semantic alignment is achieved through supervision from matching image-text pairs. To show the benefits of our semi-supervised representation, we apply it to cross-modal retrieval and phrase localization. We not only achieve state-of-the-art accuracy, but significantly better generalization across datasets, owing to the semantic continuity of the latent space.

Anne S. Wannenwetsch, Martin Kiefel, Peter V. Gehler et al.

Dense prediction tasks typically employ encoder-decoder architectures, but the prevalent convolutions in the decoder are not image-adaptive and can lead to boundary artifacts. Different generalized convolution operations have been introduced to counteract this. We go beyond these by leveraging guidance data to redefine their inherent notion of proximity. Our proposed network layer builds on the permutohedral lattice, which performs sparse convolutions in a high-dimensional space allowing for powerful non-local operations despite small filters. Multiple features with different characteristics span this permutohedral space. In contrast to prior work, we learn these features in a task-specific manner by generalizing the basic permutohedral operations to learnt feature representations. As the resulting objective is complex, a carefully designed framework and learning procedure are introduced, yielding rich feature embeddings in practice. We demonstrate the general applicability of our approach in different joint upsampling tasks. When adding our network layer to state-of-the-art networks for optical flow and semantic segmentation, boundary artifacts are removed and the accuracy is improved.

Patrick Dendorfer, Hamid Rezatofighi, Anton Milan et al.

Standardized benchmarks are crucial for the majority of computer vision applications. Although leaderboards and ranking tables should not be over-claimed, benchmarks often provide the most objective measure of performance and are therefore important guides for research. The benchmark for Multiple Object Tracking, MOTChallenge, was launched with the goal to establish a standardized evaluation of multiple object tracking methods. The challenge focuses on multiple people tracking, since pedestrians are well studied in the tracking community, and precise tracking and detection has high practical relevance. Since the first release, MOT15, MOT16 and MOT17 have tremendously contributed to the community by introducing a clean dataset and precise framework to benchmark multi-object trackers. In this paper, we present our CVPR19 benchmark, consisting of 8 new sequences depicting very crowded challenging scenes. The benchmark will be presented at the 4th BMTT MOT Challenge Workshop at the Computer Vision and Pattern Recognition Conference (CVPR) 2019, and will evaluate the state-of-the-art in multiple object tracking whend handling extremely crowded scenarios.

Junhwa Hur, Stefan Roth

Deep learning approaches to optical flow estimation have seen rapid progress over the recent years. One common trait of many networks is that they refine an initial flow estimate either through multiple stages or across the levels of a coarse-to-fine representation. While leading to more accurate results, the downside of this is an increased number of parameters. Taking inspiration from both classical energy minimization approaches as well as residual networks, we propose an iterative residual refinement (IRR) scheme based on weight sharing that can be combined with several backbone networks. It reduces the number of parameters, improves the accuracy, or even achieves both. Moreover, we show that integrating occlusion prediction and bi-directional flow estimation into our IRR scheme can further boost the accuracy. Our full network achieves state-of-the-art results for both optical flow and occlusion estimation across several standard datasets.

Nikita Araslanov, Constantin Rothkopf, Stefan Roth

Most approaches to visual scene analysis have emphasised parallel processing of the image elements. However, one area in which the sequential nature of vision is apparent, is that of segmenting multiple, potentially similar and partially occluded objects in a scene. In this work, we revisit the recurrent formulation of this challenging problem in the context of reinforcement learning. Motivated by the limitations of the global max-matching assignment of the ground-truth segments to the recurrent states, we develop an actor-critic approach in which the actor recurrently predicts one instance mask at a time and utilises the gradient from a concurrently trained critic network. We formulate the state, action, and the reward such as to let the critic model long-term effects of the current prediction and incorporate this information into the gradient signal. Furthermore, to enable effective exploration in the inherently high-dimensional action space of instance masks, we learn a compact representation using a conditional variational auto-encoder. We show that our actor-critic model consistently provides accuracy benefits over the recurrent baseline on standard instance segmentation benchmarks.

Tobias Plötz, Stefan Roth

Non-local methods exploiting the self-similarity of natural signals have been well studied, for example in image analysis and restoration. Existing approaches, however, rely on k-nearest neighbors (KNN) matching in a fixed feature space. The main hurdle in optimizing this feature space w.r.t. application performance is the non-differentiability of the KNN selection rule. To overcome this, we propose a continuous deterministic relaxation of KNN selection that maintains differentiability w.r.t. pairwise distances, but retains the original KNN as the limit of a temperature parameter approaching zero. To exploit our relaxation, we propose the neural nearest neighbors block (N3 block), a novel non-local processing layer that leverages the principle of self-similarity and can be used as building block in modern neural network architectures. We show its effectiveness for the set reasoning task of correspondence classification as well as for image restoration, including image denoising and single image super-resolution, where we outperform strong convolutional neural network (CNN) baselines and recent non-local models that rely on KNN selection in hand-chosen features spaces.

Jan-Martin O. Steitz, Faraz Saeedan, Stefan Roth

Motivated by the detection of prohibited objects in carry-on luggage as a part of avionic security screening, we develop a CNN-based object detection approach for multi-view X-ray image data. Our contributions are two-fold. First, we introduce a novel multi-view pooling layer to perform a 3D aggregation of 2D CNN-features extracted from each view. To that end, our pooling layer exploits the known geometry of the imaging system to ensure geometric consistency of the feature aggregation. Second, we introduce an end-to-end trainable multi-view detection pipeline based on Faster R-CNN, which derives the region proposals and performs the final classification in 3D using these aggregated multi-view features. Our approach shows significant accuracy gains compared to single-view detection while even being more efficient than performing single-view detection in each view.

Jochen Gast, Stefan Roth

Even though probabilistic treatments of neural networks have a long history, they have not found widespread use in practice. Sampling approaches are often too slow already for simple networks. The size of the inputs and the depth of typical CNN architectures in computer vision only compound this problem. Uncertainty in neural networks has thus been largely ignored in practice, despite the fact that it may provide important information about the reliability of predictions and the inner workings of the network. In this paper, we introduce two lightweight approaches to making supervised learning with probabilistic deep networks practical: First, we suggest probabilistic output layers for classification and regression that require only minimal changes to existing networks. Second, we employ assumed density filtering and show that activation uncertainties can be propagated in a practical fashion through the entire network, again with minor changes. Both probabilistic networks retain the predictive power of the deterministic counterpart, but yield uncertainties that correlate well with the empirical error induced by their predictions. Moreover, the robustness to adversarial examples is significantly increased.

Stephan R. Richter, Stefan Roth

In this paper, we develop novel, efficient 2D encodings for 3D geometry, which enable reconstructing full 3D shapes from a single image at high resolution. The key idea is to pose 3D shape reconstruction as a 2D prediction problem. To that end, we first develop a simple baseline network that predicts entire voxel tubes at each pixel of a reference view. By leveraging well-proven architectures for 2D pixel-prediction tasks, we attain state-of-the-art results, clearly outperforming purely voxel-based approaches. We scale this baseline to higher resolutions by proposing a memory-efficient shape encoding, which recursively decomposes a 3D shape into nested shape layers, similar to the pieces of a Matryoshka doll. This allows reconstructing highly detailed shapes with complex topology, as demonstrated in extensive experiments; we clearly outperform previous octree-based approaches despite having a much simpler architecture using standard network components. Our Matryoshka networks further enable reconstructing shapes from IDs or shape similarity, as well as shape sampling.

Faraz Saeedan, Nicolas Weber, Michael Goesele et al.

Most convolutional neural networks use some method for gradually downscaling the size of the hidden layers. This is commonly referred to as pooling, and is applied to reduce the number of parameters, improve invariance to certain distortions, and increase the receptive field size. Since pooling by nature is a lossy process, it is crucial that each such layer maintains the portion of the activations that is most important for the network's discriminability. Yet, simple maximization or averaging over blocks, max or average pooling, or plain downsampling in the form of strided convolutions are the standard. In this paper, we aim to leverage recent results on image downscaling for the purposes of deep learning. Inspired by the human visual system, which focuses on local spatial changes, we propose detail-preserving pooling (DPP), an adaptive pooling method that magnifies spatial changes and preserves important structural detail. Importantly, its parameters can be learned jointly with the rest of the network. We analyze some of its theoretical properties and show its empirical benefits on several datasets and networks, where DPP consistently outperforms previous pooling approaches.

Tobias Plötz, Anne S. Wannenwetsch, Stefan Roth

Variational inference has experienced a recent surge in popularity owing to stochastic approaches, which have yielded practical tools for a wide range of model classes. A key benefit is that stochastic variational inference obviates the tedious process of deriving analytical expressions for closed-form variable updates. Instead, one simply needs to derive the gradient of the log-posterior, which is often much easier. Yet for certain model classes, the log-posterior itself is difficult to optimize using standard gradient techniques. One such example are random field models, where optimization based on gradient linearization has proven popular, since it speeds up convergence significantly and can avoid poor local optima. In this paper we propose stochastic variational inference with gradient linearization (SVIGL). It is similarly convenient as standard stochastic variational inference - all that is required is a local linearization of the energy gradient. Its benefit over stochastic variational inference with conventional gradient methods is a clear improvement in convergence speed, while yielding comparable or even better variational approximations in terms of KL divergence. We demonstrate the benefits of SVIGL in three applications: Optical flow estimation, Poisson-Gaussian denoising, and 3D surface reconstruction.

Simon Meister, Junhwa Hur, Stefan Roth

In the era of end-to-end deep learning, many advances in computer vision are driven by large amounts of labeled data. In the optical flow setting, however, obtaining dense per-pixel ground truth for real scenes is difficult and thus such data is rare. Therefore, recent end-to-end convolutional networks for optical flow rely on synthetic datasets for supervision, but the domain mismatch between training and test scenarios continues to be a challenge. Inspired by classical energy-based optical flow methods, we design an unsupervised loss based on occlusion-aware bidirectional flow estimation and the robust census transform to circumvent the need for ground truth flow. On the KITTI benchmarks, our unsupervised approach outperforms previous unsupervised deep networks by a large margin, and is even more accurate than similar supervised methods trained on synthetic datasets alone. By optionally fine-tuning on the KITTI training data, our method achieves competitive optical flow accuracy on the KITTI 2012 and 2015 benchmarks, thus in addition enabling generic pre-training of supervised networks for datasets with limited amounts of ground truth.

Anne S. Wannenwetsch, Margret Keuper, Stefan Roth

Optical flow estimation remains challenging due to untextured areas, motion boundaries, occlusions, and more. Thus, the estimated flow is not equally reliable across the image. To that end, post-hoc confidence measures have been introduced to assess the per-pixel reliability of the flow. We overcome the artificial separation of optical flow and confidence estimation by introducing a method that jointly predicts optical flow and its underlying uncertainty. Starting from common energy-based formulations, we rely on the corresponding posterior distribution of the flow given the images. We derive a variational inference scheme based on mean field, which incorporates best practices from energy minimization. An uncertainty measure is obtained along the flow at every pixel as the (marginal) entropy of the variational distribution. We demonstrate the flexibility of our probabilistic approach by applying it to two different energies and on two benchmarks. We not only obtain flow results that are competitive with the underlying energy minimization approach, but also a reliable uncertainty measure that significantly outperforms existing post-hoc approaches.

Junhwa Hur, Stefan Roth

Optical flow estimation is one of the most studied problems in computer vision, yet recent benchmark datasets continue to reveal problem areas of today's approaches. Occlusions have remained one of the key challenges. In this paper, we propose a symmetric optical flow method to address the well-known chicken-and-egg relation between optical flow and occlusions. In contrast to many state-of-the-art methods that consider occlusions as outliers, possibly filtered out during post-processing, we highlight the importance of joint occlusion reasoning in the optimization and show how to utilize occlusion as an important cue for estimating optical flow. The key feature of our model is to fully exploit the symmetry properties that characterize optical flow and occlusions in the two consecutive images. Specifically through utilizing forward-backward consistency and occlusion-disocclusion symmetry in the energy, our model jointly estimates optical flow in both forward and backward direction, as well as consistent occlusion maps in both views. We demonstrate significant performance benefits on standard benchmarks, especially from the occlusion-disocclusion symmetry. On the challenging KITTI dataset we report the most accurate two-frame results to date.

Florian Lang, Tobias Plötz, Stefan Roth

Photographing scenes with high dynamic range (HDR) poses great challenges to consumer cameras with their limited sensor bit depth. To address this, Zhao et al. recently proposed a novel sensor concept - the modulo camera - which captures the least significant bits of the recorded scene instead of going into saturation. Similar to conventional pipelines, HDR images can be reconstructed from multiple exposures, but significantly fewer images are needed than with a typical saturating sensor. While the concept is appealing, we show that the original reconstruction approach assumes noise-free measurements and quickly breaks down otherwise. To address this, we propose a novel reconstruction algorithm that is robust to image noise and produces significantly fewer artifacts. We theoretically analyze correctness as well as limitations, and show that our approach significantly outperforms the baseline on real data.

Tobias Plötz, Stefan Roth

Lacking realistic ground truth data, image denoising techniques are traditionally evaluated on images corrupted by synthesized i.i.d. Gaussian noise. We aim to obviate this unrealistic setting by developing a methodology for benchmarking denoising techniques on real photographs. We capture pairs of images with different ISO values and appropriately adjusted exposure times, where the nearly noise-free low-ISO image serves as reference. To derive the ground truth, careful post-processing is needed. We correct spatial misalignment, cope with inaccuracies in the exposure parameters through a linear intensity transform based on a novel heteroscedastic Tobit regression model, and remove residual low-frequency bias that stems, e.g., from minor illumination changes. We then capture a novel benchmark dataset, the Darmstadt Noise Dataset (DND), with consumer cameras of differing sensor sizes. One interesting finding is that various recent techniques that perform well on synthetic noise are clearly outperformed by BM3D on photographs with real noise. Our benchmark delineates realistic evaluation scenarios that deviate strongly from those commonly used in the scientific literature.

Laura Leal-Taixé, Anton Milan, Konrad Schindler et al.

Standardized benchmarks are crucial for the majority of computer vision applications. Although leaderboards and ranking tables should not be over-claimed, benchmarks often provide the most objective measure of performance and are therefore important guides for research. We present a benchmark for Multiple Object Tracking launched in the late 2014, with the goal of creating a framework for the standardized evaluation of multiple object tracking methods. This paper collects the two releases of the benchmark made so far, and provides an in-depth analysis of almost 50 state-of-the-art trackers that were tested on over 11000 frames. We show the current trends and weaknesses of multiple people tracking methods, and provide pointers of what researchers should be focusing on to push the field forward.

Marius Cordts, Timo Rehfeld, Lukas Schneider et al.

Recent progress in advanced driver assistance systems and the race towards autonomous vehicles is mainly driven by two factors: (1) increasingly sophisticated algorithms that interpret the environment around the vehicle and react accordingly, and (2) the continuous improvements of sensor technology itself. In terms of cameras, these improvements typically include higher spatial resolution, which as a consequence requires more data to be processed. The trend to add multiple cameras to cover the entire surrounding of the vehicle is not conducive in that matter. At the same time, an increasing number of special purpose algorithms need access to the sensor input data to correctly interpret the various complex situations that can occur, particularly in urban traffic. By observing those trends, it becomes clear that a key challenge for vision architectures in intelligent vehicles is to share computational resources. We believe this challenge should be faced by introducing a representation of the sensory data that provides compressed and structured access to all relevant visual content of the scene. The Stixel World discussed in this paper is such a representation. It is a medium-level model of the environment that is specifically designed to compress information about obstacles by leveraging the typical layout of outdoor traffic scenes. It has proven useful for a multitude of automotive vision applications, including object detection, tracking, segmentation, and mapping. In this paper, we summarize the ideas behind the model and generalize it to take into account multiple dense input streams: the image itself, stereo depth maps, and semantic class probability maps that can be generated, e.g., by CNNs. Our generalization is embedded into a novel mathematical formulation for the Stixel model. We further sketch how the free parameters of the model can be learned using structured SVMs.