Search

In this paper, we present a novel architecture for interactive segmentation in winter sports contexts. The field of interactive segmentation deals with the prediction of high-quality segmentation masks by informing the network about the objects position with the help of user guidance. In our case the guidance consists of click prompts. For this task, we first present a baseline architecture which is specifically geared towards quickly responding after each click. Afterwards, we motivate and describe a number of architectural modifications which improve the performance when tasked with segmenting winter sports equipment on the WSESeg dataset. With regards to the average NoC@85 metric on the WSESeg classes, we outperform SAM and HQ-SAM by 2.336 and 7.946 clicks, respectively. When applied to the HQSeg-44k dataset, our system delivers state-of-the-art results with a NoC@90 of 6.00 and NoC@95 of 9.89. In addition to that, we test our model on a novel dataset containing masks for humans during skiing., Comment: 4 figures, 6 tables, 12 pages

Assistive technology can be leveraged by blind people when searching for objects in their daily lives. We created ObjectFinder, an open-vocabulary interactive object-search prototype, which combines object detection with scene description and navigation. It enables blind persons to detect and navigate to objects of their choice. Our approach used co-design for the development of the prototype. We further conducted need-finding interviews to better understand challenges in object search, followed by a study with the ObjectFinder prototype in a laboratory setting simulating a living room and an office, with eight blind users. Additionally, we compared the prototype with BeMyEyes and Lookout for object search. We found that most participants felt more independent with ObjectFinder and preferred it over the baselines when deployed on more efficient hardware, as it enhances mental mapping and allows for active target definition. Moreover, we identified factors for future directions for the development of object-search systems.

Monocular geometric scene understanding combines panoptic segmentation and self-supervised depth estimation, focusing on real-time application in autonomous vehicles. We introduce MGNiceNet, a unified approach that uses a linked kernel formulation for panoptic segmentation and self-supervised depth estimation. MGNiceNet is based on the state-of-the-art real-time panoptic segmentation method RT-K-Net and extends the architecture to cover both panoptic segmentation and self-supervised monocular depth estimation. To this end, we introduce a tightly coupled self-supervised depth estimation predictor that explicitly uses information from the panoptic path for depth prediction. Furthermore, we introduce a panoptic-guided motion masking method to improve depth estimation without relying on video panoptic segmentation annotations. We evaluate our method on two popular autonomous driving datasets, Cityscapes and KITTI. Our model shows state-of-the-art results compared to other real-time methods and closes the gap to computationally more demanding methods. Source code and trained models are available at https://github.com/markusschoen/MGNiceNet., Comment: Accepted for ACCV 2024

Depth estimation is an essential task toward full scene understanding since it allows the projection of rich semantic information captured by cameras into 3D space. While the field has gained much attention recently, datasets for depth estimation lack scene diversity or sensor modalities. This work presents the ADUULM-360 dataset, a novel multi-modal dataset for depth estimation. The ADUULM-360 dataset covers all established autonomous driving sensor modalities, cameras, lidars, and radars. It covers a frontal-facing stereo setup, six surround cameras covering the full 360-degree, two high-resolution long-range lidar sensors, and five long-range radar sensors. It is also the first depth estimation dataset that contains diverse scenes in good and adverse weather conditions. We conduct extensive experiments using state-of-the-art self-supervised depth estimation methods under different training tasks, such as monocular training, stereo training, and full surround training. Discussing these results, we demonstrate common limitations of state-of-the-art methods, especially in adverse weather conditions, which hopefully will inspire future research in this area. Our dataset, development kit, and trained baselines are available at https://github.com/uulm-mrm/aduulm_360_dataset., Comment: 2024 IEEE International Conference on Intelligent Transportation Systems (ITSC)

For the validation and verification of automotive radars, datasets of realistic traffic scenarios are required, which, how ever, are laborious to acquire. In this paper, we introduce radar scene synthesis using GANs as an alternative to the real dataset acquisition and simulation-based approaches. We train a PointNet++ based GAN model to generate realistic radar point cloud scenes and use a binary classifier to evaluate the performance of scenes generated using this model against a test set of real scenes. We demonstrate that our GAN model achieves similar performance (~87%) to the real scenes test set., Comment: ICMIM 2024; 7th IEEE MTT Conference

Image monitoring and guidance during medical examinations can aid both diagnosis and treatment. However, the sampling frequency is often too low, which creates a need to estimate the missing images. We present a probabilistic motion model for sequential medical images, with the ability to both estimate motion between acquired images and forecast the motion ahead of time. The core is a low-dimensional temporal process based on a linear Gaussian state-space model with analytically tractable solutions for forecasting, simulation, and imputation of missing samples. The results, from two experiments on publicly available cardiac datasets, show reliable motion estimates and an improved forecasting performance using patient-specific adaptation by online learning., Comment: Medical Image Computing and Computer Assisted Intervention (MICCAI) 2024

Diffusion models have achieved remarkable progress in generative modelling, particularly in enhancing image quality to conform to human preferences. Recently, these models have also been applied to low-level computer vision for photo-realistic image restoration (IR) in tasks such as image denoising, deblurring, dehazing, etc. In this review paper, we introduce key constructions in diffusion models and survey contemporary techniques that make use of diffusion models in solving general IR tasks. Furthermore, we point out the main challenges and limitations of existing diffusion-based IR frameworks and provide potential directions for future work., Comment: Review paper; any comments and suggestions are most welcome!

In this paper we introduce a new dataset containing instance segmentation masks for ten different categories of winter sports equipment, called WSESeg (Winter Sports Equipment Segmentation). Furthermore, we carry out interactive segmentation experiments on said dataset to explore possibilities for efficient further labeling. The SAM and HQ-SAM models are conceptualized as foundation models for performing user guided segmentation. In order to measure their claimed generalization capability we evaluate them on WSESeg. Since interactive segmentation offers the benefit of creating easily exploitable ground truth data during test-time, we are going to test various online adaptation methods for the purpose of exploring potentials for improvements without having to fine-tune the models explicitly. Our experiments show that our adaptation methods drastically reduce the Failure Rate (FR) and Number of Clicks (NoC) metrics, which generally leads faster to better interactive segmentation results., Comment: 7 pages, 1 figure, 3 tables, Accepted at CBMI 2024

In many manufacturing settings, annotating data for machine learning and computer vision is costly, but synthetic data can be generated at significantly lower cost. Substituting the real-world data with synthetic data is therefore appealing for many machine learning applications that require large amounts of training data. However, relying solely on synthetic data is frequently inadequate for effectively training models that perform well on real-world data, primarily due to domain shifts between the synthetic and real-world data. We discuss approaches for dealing with such a domain shift when detecting defects in X-ray scans of aluminium wheels. Using both simulated and real-world X-ray images, we train an object detection model with different strategies to identify the training approach that generates the best detection results while minimising the demand for annotated real-world training samples. Our preliminary findings suggest that the sim-2-real domain adaptation approach is more cost-efficient than a fully supervised oracle - if the total number of available annotated samples is fixed. Given a certain number of labeled real-world samples, training on a mix of synthetic and unlabeled real-world data achieved comparable or even better detection results at significantly lower cost. We argue that future research into the cost-efficiency of different training strategies is important for a better understanding of how to allocate budget in applied machine learning projects.

Utilizing transformer architectures for semantic segmentation of high-resolution images is hindered by the attention's quadratic computational complexity in the number of tokens. A solution to this challenge involves decreasing the number of tokens through token merging, which has exhibited remarkable enhancements in inference speed, training efficiency, and memory utilization for image classification tasks. In this paper, we explore various token merging strategies within the framework of the Segformer architecture and perform experiments on multiple semantic segmentation and human pose estimation datasets. Notably, without model re-training, we, for example, achieve an inference acceleration of 61% on the Cityscapes dataset while maintaining the mIoU performance. Consequently, this paper facilitates the deployment of transformer-based architectures on resource-constrained devices and in real-time applications., Comment: 7 pages, to be published in IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR) 2024

Though diffusion models have been successfully applied to various image restoration (IR) tasks, their performance is sensitive to the choice of training datasets. Typically, diffusion models trained in specific datasets fail to recover images that have out-of-distribution degradations. To address this problem, this work leverages a capable vision-language model and a synthetic degradation pipeline to learn image restoration in the wild (wild IR). More specifically, all low-quality images are simulated with a synthetic degradation pipeline that contains multiple common degradations such as blur, resize, noise, and JPEG compression. Then we introduce robust training for a degradation-aware CLIP model to extract enriched image content features to assist high-quality image restoration. Our base diffusion model is the image restoration SDE (IR-SDE). Built upon it, we further present a posterior sampling strategy for fast noise-free image generation. We evaluate our model on both synthetic and real-world degradation datasets. Moreover, experiments on the unified image restoration task illustrate that the proposed posterior sampling improves image generation quality for various degradations., Comment: CVPRW 2024; Code: https://github.com/Algolzw/daclip-uir

Scene graph generation has emerged as a prominent research field in computer vision, witnessing significant advancements in the recent years. However, despite these strides, precise and thorough definitions for the metrics used to evaluate scene graph generation models are lacking. In this paper, we address this gap in the literature by providing a review and precise definition of commonly used metrics in scene graph generation. Our comprehensive examination clarifies the underlying principles of these metrics and can serve as a reference or introduction to scene graph metrics. Furthermore, to facilitate the usage of these metrics, we introduce a standalone Python package called SGBench that efficiently implements all defined metrics, ensuring their accessibility to the research community. Additionally, we present a scene graph benchmarking web service, that enables researchers to compare scene graph generation methods and increase visibility of new methods in a central place. All of our code can be found at https://lorjul.github.io/sgbench/.

The interactive segmentation task consists in the creation of object segmentation masks based on user interactions. The most common way to guide a model towards producing a correct segmentation consists in clicks on the object and background. The recently published Segment Anything Model (SAM) supports a generalized version of the interactive segmentation problem and has been trained on an object segmentation dataset which contains 1.1B masks. Though being trained extensively and with the explicit purpose of serving as a foundation model, we show significant limitations of SAM when being applied for interactive segmentation on novel domains or object types. On the used datasets, SAM displays a failure rate $\text{FR}_{30}@90$ of up to $72.6 \%$. Since we still want such foundation models to be immediately applicable, we present a framework that can adapt SAM during immediate usage. For this we will leverage the user interactions and masks, which are constructed during the interactive segmentation process. We use this information to generate pseudo-labels, which we use to compute a loss function and optimize a part of the SAM model. The presented method causes a relative reduction of up to $48.1 \%$ in the $\text{FR}_{20}@85$ and $46.6 \%$ in the $\text{FR}_{30}@90$ metrics., Comment: 11 pages, 2 figures, 4 tables

Purpose. To present SPINEPS, an open-source deep learning approach for semantic and instance segmentation of 14 spinal structures (ten vertebra substructures, intervertebral discs, spinal cord, spinal canal, and sacrum) in whole body T2w MRI. Methods. During this HIPPA-compliant, retrospective study, we utilized the public SPIDER dataset (218 subjects, 63% female) and a subset of the German National Cohort (1423 subjects, mean age 53, 49% female) for training and evaluation. We combined CT and T2w segmentations to train models that segment 14 spinal structures in T2w sagittal scans both semantically and instance-wise. Performance evaluation metrics included Dice similarity coefficient, average symmetrical surface distance, panoptic quality, segmentation quality, and recognition quality. Statistical significance was assessed using the Wilcoxon signed-rank test. An in-house dataset was used to qualitatively evaluate out-of-distribution samples. Results. On the public dataset, our approach outperformed the baseline (instance-wise vertebra dice score 0.929 vs. 0.907, p-value<0.001). Training on auto-generated annotations and evaluating on manually corrected test data from the GNC yielded global dice scores of 0.900 for vertebrae, 0.960 for intervertebral discs, and 0.947 for the spinal canal. Incorporating the SPIDER dataset during training increased these scores to 0.920, 0.967, 0.958, respectively. Conclusions. The proposed segmentation approach offers robust segmentation of 14 spinal structures in T2w sagittal images, including the spinal cord, spinal canal, intervertebral discs, endplate, sacrum, and vertebrae. The approach yields both a semantic and instance mask as output, thus being easy to utilize. This marks the first publicly available algorithm for whole spine segmentation in sagittal T2w MR imaging., Comment: https://github.com/Hendrik-code/spineps

Accurate and comprehensive semantic segmentation of Bird's Eye View (BEV) is essential for ensuring safe and proactive navigation in autonomous driving. Although cooperative perception has exceeded the detection capabilities of single-agent systems, prevalent camera-based algorithms in cooperative perception neglect valuable information derived from historical observations. This limitation becomes critical during sensor failures or communication issues as cooperative perception reverts to single-agent perception, leading to degraded performance and incomplete BEV segmentation maps. This paper introduces TempCoBEV, a temporal module designed to incorporate historical cues into current observations, thereby improving the quality and reliability of BEV map segmentations. We propose an importance-guided attention architecture to effectively integrate temporal information that prioritizes relevant properties for BEV map segmentation. TempCoBEV is an independent temporal module that seamlessly integrates into state-of-the-art camera-based cooperative perception models. We demonstrate through extensive experiments on the OPV2V dataset that TempCoBEV performs better than non-temporal models in predicting current and future BEV map segmentations, particularly in scenarios involving communication failures. We show the efficacy of TempCoBEV and its capability to integrate historical cues into the current BEV map, improving predictions under optimal communication conditions by up to 2% and under communication failures by up to 19%. The code will be published on GitHub.

Vision-language models such as CLIP have shown great impact on diverse downstream tasks for zero-shot or label-free predictions. However, when it comes to low-level vision such as image restoration their performance deteriorates dramatically due to corrupted inputs. In this paper, we present a degradation-aware vision-language model (DA-CLIP) to better transfer pretrained vision-language models to low-level vision tasks as a multi-task framework for image restoration. More specifically, DA-CLIP trains an additional controller that adapts the fixed CLIP image encoder to predict high-quality feature embeddings. By integrating the embedding into an image restoration network via cross-attention, we are able to pilot the model to learn a high-fidelity image reconstruction. The controller itself will also output a degradation feature that matches the real corruptions of the input, yielding a natural classifier for different degradation types. In addition, we construct a mixed degradation dataset with synthetic captions for DA-CLIP training. Our approach advances state-of-the-art performance on both \emph{degradation-specific} and \emph{unified} image restoration tasks, showing a promising direction of prompting image restoration with large-scale pretrained vision-language models. Our code is available at https://github.com/Algolzw/daclip-uir., Comment: Accepted by ICLR 2024. Project page: https://algolzw.github.io/daclip-uir/index.html

The main approaches for simulating FMCW radar are based on ray tracing, which is usually computationally intensive and do not account for background noise. This work proposes a faster method for FMCW radar simulation capable of generating synthetic raw radar data using generative adversarial networks (GAN). The code and pre-trained weights are open-source and available on GitHub. This method generates 16 simultaneous chirps, which allows the generated data to be used for the further development of algorithms for processing radar data (filtering and clustering). This can increase the potential for data augmentation, e.g., by generating data in non-existent or safety-critical scenarios that are not reproducible in real life. In this work, the GAN was trained with radar measurements of a motorcycle and used to generate synthetic raw radar data of a motorcycle traveling in a straight line. For generating this data, the distance of the motorcycle and Gaussian noise are used as input to the neural network. The synthetic generated radar chirps were evaluated using the Frechet Inception Distance (FID). Then, the Range-Azimuth (RA) map is calculated twice: first, based on synthetic data using this GAN and, second, based on real data. Based on these RA maps, an algorithm with adaptive threshold and edge detection is used for object detection. The results have shown that the data is realistic in terms of coherent radar reflections of the motorcycle and background noise based on the comparison of chirps, the RA maps and the object detection results. Thus, the proposed method in this work has shown to minimize the simulation-to-reality gap for the generation of radar data.

Panoptic segmentation is one of the most challenging scene parsing tasks, combining the tasks of semantic segmentation and instance segmentation. While much progress has been made, few works focus on the real-time application of panoptic segmentation methods. In this paper, we revisit the recently introduced K-Net architecture. We propose vital changes to the architecture, training, and inference procedure, which massively decrease latency and improve performance. Our resulting RT-K-Net sets a new state-of-the-art performance for real-time panoptic segmentation methods on the Cityscapes dataset and shows promising results on the challenging Mapillary Vistas dataset. On Cityscapes, RT-K-Net reaches 60.2 % PQ with an average inference time of 32 ms for full resolution 1024x2048 pixel images on a single Titan RTX GPU. On Mapillary Vistas, RT-K-Net reaches 33.2 % PQ with an average inference time of 69 ms. Source code is available at https://github.com/markusschoen/RT-K-Net.

This work aims to improve the applicability of diffusion models in realistic image restoration. Specifically, we enhance the diffusion model in several aspects such as network architecture, noise level, denoising steps, training image size, and optimizer/scheduler. We show that tuning these hyperparameters allows us to achieve better performance on both distortion and perceptual scores. We also propose a U-Net based latent diffusion model which performs diffusion in a low-resolution latent space while preserving high-resolution information from the original input for the decoding process. Compared to the previous latent-diffusion model which trains a VAE-GAN to compress the image, our proposed U-Net compression strategy is significantly more stable and can recover highly accurate images without relying on adversarial optimization. Importantly, these modifications allow us to apply diffusion models to various image restoration tasks, including real-world shadow removal, HR non-homogeneous dehazing, stereo super-resolution, and bokeh effect transformation. By simply replacing the datasets and slightly changing the noise network, our model, named Refusion, is able to deal with large-size images (e.g., 6000 x 4000 x 3 in HR dehazing) and produces good results on all the above restoration problems. Our Refusion achieves the best perceptual performance in the NTIRE 2023 Image Shadow Removal Challenge and wins 2nd place overall., Comment: CVPRW 2023. Runner-up method in NTIRE 2023 Image Shadow Removal Challenge. Code is available at https://github.com/Algolzw/image-restoration-sde

Pseudo depth maps are depth map predicitions which are used as ground truth during training. In this paper we leverage pseudo depth maps in order to segment objects of classes that have never been seen during training. This renders our object segmentation task an open world task. The pseudo depth maps are generated using pretrained networks, which have either been trained with the full intention to generalize to downstream tasks (LeRes and MiDaS), or which have been trained in an unsupervised fashion on video sequences (MonodepthV2). In order to tell our network which object to segment, we provide the network with a single click on the object's surface on the pseudo depth map of the image as input. We test our approach on two different scenarios: One without the RGB image and one where the RGB image is part of the input. Our results demonstrate a considerably better generalization performance from seen to unseen object types when depth is used. On the Semantic Boundaries Dataset we achieve an improvement from $61.57$ to $69.79$ IoU score on unseen classes, when only using half of the training classes during training and performing the segmentation on depth maps only., Comment: Accepted to L3D-IVU Workshop at CVPR 2023

Performance analyses based on videos are commonly used by coaches of athletes in various sports disciplines. In individual sports, these analyses mainly comprise the body posture. This paper focuses on the disciplines of triple, high, and long jump, which require fine-grained locations of the athlete's body. Typical human pose estimation datasets provide only a very limited set of keypoints, which is not sufficient in this case. Therefore, we propose a method to detect arbitrary keypoints on the whole body of the athlete by leveraging the limited set of annotated keypoints and auto-generated segmentation masks of body parts. Evaluations show that our model is capable of detecting keypoints on the head, torso, hands, feet, arms, and legs, including also bent elbows and knees. We analyze and compare different techniques to encode desired keypoints as the model's input and their embedding for the Transformer backbone., Comment: Accepted at CVSports23 (Workshop at CVPR 23)

Many important computer vision applications are naturally formulated as regression problems. Within medical imaging, accurate regression models have the potential to automate various tasks, helping to lower costs and improve patient outcomes. Such safety-critical deployment does however require reliable estimation of model uncertainty, also under the wide variety of distribution shifts that might be encountered in practice. Motivated by this, we set out to investigate the reliability of regression uncertainty estimation methods under various real-world distribution shifts. To that end, we propose an extensive benchmark of 8 image-based regression datasets with different types of challenging distribution shifts. We then employ our benchmark to evaluate many of the most common uncertainty estimation methods, as well as two state-of-the-art uncertainty scores from the task of out-of-distribution detection. We find that while methods are well calibrated when there is no distribution shift, they all become highly overconfident on many of the benchmark datasets. This uncovers important limitations of current uncertainty estimation methods, and the proposed benchmark therefore serves as a challenge to the research community. We hope that our benchmark will spur more work on how to develop truly reliable regression uncertainty estimation methods. Code is available at https://github.com/fregu856/regression_uncertainty., Comment: TMLR, 2023. Code is available at https://github.com/fregu856/regression_uncertainty

This paper presents a stochastic differential equation (SDE) approach for general-purpose image restoration. The key construction consists in a mean-reverting SDE that transforms a high-quality image into a degraded counterpart as a mean state with fixed Gaussian noise. Then, by simulating the corresponding reverse-time SDE, we are able to restore the origin of the low-quality image without relying on any task-specific prior knowledge. Crucially, the proposed mean-reverting SDE has a closed-form solution, allowing us to compute the ground truth time-dependent score and learn it with a neural network. Moreover, we propose a maximum likelihood objective to learn an optimal reverse trajectory that stabilizes the training and improves the restoration results. The experiments show that our proposed method achieves highly competitive performance in quantitative comparisons on image deraining, deblurring, and denoising, setting a new state-of-the-art on two deraining datasets. Finally, the general applicability of our approach is further demonstrated via qualitative results on image super-resolution, inpainting, and dehazing. Code is available at https://github.com/Algolzw/image-restoration-sde., Comment: Accepted by ICML 2023; Project page: https://algolzw.github.io/ir-sde/index.html

Medical imaging plays a vital role in modern diagnostics and treatment. The temporal nature of disease or treatment progression often results in longitudinal data. Due to the cost and potential harm, acquiring large medical datasets necessary for deep learning can be difficult. Medical image synthesis could help mitigate this problem. However, until now, the availability of GANs capable of synthesizing longitudinal volumetric data has been limited. To address this, we use the recent advances in latent space-based image editing to propose a novel joint learning scheme to explicitly embed temporal dependencies in the latent space of GANs. This, in contrast to previous methods, allows us to synthesize continuous, smooth, and high-quality longitudinal volumetric data with limited supervision. We show the effectiveness of our approach on three datasets containing different longitudinal dependencies. Namely, modeling a simple image transformation, breathing motion, and tumor regression, all while showing minimal disentanglement. The implementation is made available online at https://github.com/julschoen/Temp-GAN.

Objective: Imbalances of the electrolyte concentration levels in the body can lead to catastrophic consequences, but accurate and accessible measurements could improve patient outcomes. While blood tests provide accurate measurements, they are invasive and the laboratory analysis can be slow or inaccessible. In contrast, an electrocardiogram (ECG) is a widely adopted tool which is quick and simple to acquire. However, the problem of estimating continuous electrolyte concentrations directly from ECGs is not well-studied. We therefore investigate if regression methods can be used for accurate ECG-based prediction of electrolyte concentrations. Methods: We explore the use of deep neural networks (DNNs) for this task. We analyze the regression performance across four electrolytes, utilizing a novel dataset containing over 290000 ECGs. For improved understanding, we also study the full spectrum from continuous predictions to binary classification of extreme concentration levels. To enhance clinical usefulness, we finally extend to a probabilistic regression approach and evaluate different uncertainty estimates. Results: We find that the performance varies significantly between different electrolytes, which is clinically justified in the interplay of electrolytes and their manifestation in the ECG. We also compare the regression accuracy with that of traditional machine learning models, demonstrating superior performance of DNNs. Conclusion: Discretization can lead to good classification performance, but does not help solve the original problem of predicting continuous concentration levels. While probabilistic regression demonstrates potential practical usefulness, the uncertainty estimates are not particularly well-calibrated. Significance: Our study is a first step towards accurate and reliable ECG-based prediction of electrolyte concentration levels., Comment: Code and trained models are available at https://github.com/philippvb/ecg-electrolyte-regression

Semi-supervised learning (SSL) can reduce the need for large labelled datasets by incorporating unlabelled data into the training. This is particularly interesting for semantic segmentation, where labelling data is very costly and time-consuming. Current SSL approaches use an initially supervised trained model to generate predictions for unlabelled images, called pseudo-labels, which are subsequently used for training a new model from scratch. Since the predictions usually do not come from an error-free neural network, they are naturally full of errors. However, training with partially incorrect labels often reduce the final model performance. Thus, it is crucial to manage errors/noise of pseudo-labels wisely. In this work, we use three mechanisms to control pseudo-label noise and errors: (1) We construct a solid base framework by mixing images with cow-patterns on unlabelled images to reduce the negative impact of wrong pseudo-labels. Nevertheless, wrong pseudo-labels still have a negative impact on the performance. Therefore, (2) we propose a simple and effective loss weighting scheme for pseudo-labels defined by the feedback of the model trained on these pseudo-labels. This allows us to soft-weight the pseudo-label training examples based on their determined confidence score during training. (3) We also study the common practice to ignore pseudo-labels with low confidence and empirically analyse the influence and effect of pseudo-labels with different confidence ranges on SSL and the contribution of pseudo-label filtering to the achievable performance gains. We show that our method performs superior to state of-the-art alternatives on various datasets. Furthermore, we show that our findings also transfer to other tasks such as human pose estimation. Our code is available at https://github.com/ChristmasFan/SSL_Denoising_Segmentation., Comment: Accepted to BMVC 2022

Generative models such as Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs) play an increasingly important role in medical image analysis. The latent spaces of these models often show semantically meaningful directions corresponding to human-interpretable image transformations. However, until now, their exploration for medical images has been limited due to the requirement of supervised data. Several methods for unsupervised discovery of interpretable directions in GAN latent spaces have shown interesting results on natural images. This work explores the potential of applying these techniques on medical images by training a GAN and a VAE on thoracic CT scans and using an unsupervised method to discover interpretable directions in the resulting latent space. We find several directions corresponding to non-trivial image transformations, such as rotation or breast size. Furthermore, the directions show that the generative models capture 3D structure despite being presented only with 2D data. The results show that unsupervised methods to discover interpretable directions in GANs generalize to VAEs and can be applied to medical images. This opens a wide array of future work using these methods in medical image analysis., Comment: Accepted for presentation at DGM4MICCAI 2022

Since COVID strongly affects the respiratory system, lung CT-scans can be used for the analysis of a patients health. We introduce a neural network for the prediction of the severity of lung damage and the detection of a COVID-infection using three-dimensional CT-data. Therefore, we adapt the recent ConvNeXt model to process three-dimensional data. Furthermore, we design and analyze different pretraining methods specifically designed to improve the models ability to handle three-dimensional CT-data. We rank 2nd in the 1st COVID19 Severity Detection Challenge and 3rd in the 2nd COVID19 Detection Challenge., Comment: 17 pages, 3 figures, informations about challenge submission

We introduce MGNet, a multi-task framework for monocular geometric scene understanding. We define monocular geometric scene understanding as the combination of two known tasks: Panoptic segmentation and self-supervised monocular depth estimation. Panoptic segmentation captures the full scene not only semantically, but also on an instance basis. Self-supervised monocular depth estimation uses geometric constraints derived from the camera measurement model in order to measure depth from monocular video sequences only. To the best of our knowledge, we are the first to propose the combination of these two tasks in one single model. Our model is designed with focus on low latency to provide fast inference in real-time on a single consumer-grade GPU. During deployment, our model produces dense 3D point clouds with instance aware semantic labels from single high-resolution camera images. We evaluate our model on two popular autonomous driving benchmarks, i.e., Cityscapes and KITTI, and show competitive performance among other real-time capable methods. Source code is available at https://github.com/markusschoen/MGNet.

Robust environment perception for autonomous vehicles is a tremendous challenge, which makes a diverse sensor set with e.g. camera, lidar and radar crucial. In the process of understanding the recorded sensor data, 3D semantic segmentation plays an important role. Therefore, this work presents a pyramid-based deep fusion architecture for lidar and camera to improve 3D semantic segmentation of traffic scenes. Individual sensor backbones extract feature maps of camera images and lidar point clouds. A novel Pyramid Fusion Backbone fuses these feature maps at different scales and combines the multimodal features in a feature pyramid to compute valuable multimodal, multi-scale features. The Pyramid Fusion Head aggregates these pyramid features and further refines them in a late fusion step, incorporating the final features of the sensor backbones. The approach is evaluated on two challenging outdoor datasets and different fusion strategies and setups are investigated. It outperforms recent range view based lidar approaches as well as all so far proposed fusion strategies and architectures., Comment: conditionally accepted at IEEE IV 2022, 7 pages, 4 figures, 5 tables

Self-Supervised learning (SSL) has become the new state-of-art in several domain classification and segmentation tasks. Of these, one popular category in SSL is distillation networks such as BYOL. This work proposes RSDnet, which applies the distillation network (BYOL) in the remote sensing (RS) domain where data is non-trivially different from natural RGB images. Since Multi-spectral (MS) and synthetic aperture radar (SAR) sensors provide varied spectral and spatial resolution information, we utilised them as an implicit augmentation to learn invariant feature embeddings. In order to learn RS based invariant features with SSL, we trained RSDnet in two ways, i.e., single channel feature learning and three channel feature learning. This work explores the usefulness of single channel feature learning from random MS and SAR bands compared to the common notion of using three or more bands. In our linear evaluation, these single channel features reached a 0.92 F1 score on the EuroSAT classification task and 59.6 mIoU on the DFC segmentation task for certain single bands. We also compared our results with ImageNet weights and showed that the RS based SSL model outperforms the supervised ImageNet based model. We further explored the usefulness of multi-modal data compared to single modality data, and it is shown that utilising MS and SAR data learn better invariant representations than utilising only MS data., Comment: Manuscript Accepted at IEEE JSTAR

Energy-based models (EBMs) have experienced a resurgence within machine learning in recent years, including as a promising alternative for probabilistic regression. However, energy-based regression requires a proposal distribution to be manually designed for training, and an initial estimate has to be provided at test-time. We address both of these issues by introducing a conceptually simple method to automatically learn an effective proposal distribution, which is parameterized by a separate network head. To this end, we derive a surprising result, leading to a unified training objective that jointly minimizes the KL divergence from the proposal to the EBM, and the negative log-likelihood of the EBM. At test-time, we can then employ importance sampling with the trained proposal to efficiently evaluate the learned EBM and produce stand-alone predictions. Furthermore, we utilize our derived training objective to learn mixture density networks (MDNs) with a jointly trained energy-based teacher, consistently outperforming conventional MDN training on four real-world regression tasks within computer vision. Code is available at https://github.com/fregu856/ebms_proposals., Comment: AISTATS 2022. Code is available at https://github.com/fregu856/ebms_proposals

Human body volume estimation from a single RGB image is a challenging problem despite minimal attention from the research community. However VolNet, an architecture leveraging 2D and 3D pose estimation, body part segmentation and volume regression extracted from a single 2D RGB image combined with the subject's body height can be used to estimate the total body volume. VolNet is designed to predict the 2D and 3D pose as well as the body part segmentation in intermediate tasks. We generated a synthetic, large-scale dataset of photo-realistic images of human bodies with a wide range of body shapes and realistic poses called SURREALvols. By using Volnet and combining multiple stacked hourglass networks together with ResNeXt, our model correctly predicted the volume in ~82% of cases with a 10% tolerance threshold. This is a considerable improvement compared to state-of-the-art solutions such as BodyNet with only a ~38% success rate.

This work tackles the problem of temporally coherent face anonymization in natural video streams.We propose JaGAN, a two-stage system starting with detecting and masking out faces with black image patches in all individual frames of the video. The second stage leverages a privacy-preserving Video Generative Adversarial Network designed to inpaint the missing image patches with artificially generated faces. Our initial experiments reveal that image based generative models are not capable of inpainting patches showing temporal coherent appearance across neighboring video frames. To address this issue we introduce a newly curated video collection, which is made publicly available for the research community along with this paper. We also introduce the Identity Invariance Score IdI as a means to quantify temporal coherency between neighboring frames., Comment: Preprint of our FG2021 submission

Obstacle avoidance is a fundamental and challenging problem for autonomous navigation of mobile robots. In this paper, we consider the problem of obstacle avoidance in simple 3D environments where the robot has to solely rely on a single monocular camera. In particular, we are interested in solving this problem without relying on localization, mapping, or planning techniques. Most of the existing work consider obstacle avoidance as two separate problems, namely obstacle detection, and control. Inspired by the recent advantages of deep reinforcement learning in Atari games and understanding highly complex situations in Go, we tackle the obstacle avoidance problem as a data-driven end-to-end deep learning approach. Our approach takes raw images as input and generates control commands as output. We show that discrete action spaces are outperforming continuous control commands in terms of expected average reward in maze-like environments. Furthermore, we show how to accelerate the learning and increase the robustness of the policy by incorporating predicted depth maps by a generative adversarial network., Comment: Accepted at 2021 IEEE International Conference on Robotics and Automation (ICRA)

Spatiotemporal imaging has applications in e.g. cardiac diagnostics, surgical guidance, and radiotherapy monitoring, In this paper, we explain the temporal motion by identifying the underlying dynamics, only based on the sequential images. Our dynamical model maps the inputs of observed high-dimensional sequential images to a low-dimensional latent space wherein a linear relationship between a hidden state process and the lower-dimensional representation of the inputs holds. For this, we use a conditional variational auto-encoder (CVAE) to nonlinearly map the higher-dimensional image to a lower-dimensional space, wherein we model the dynamics with a linear Gaussian state-space model (LG-SSM). The model, a modified version of the Kalman variational auto-encoder, is end-to-end trainable, and the weights, both in the CVAE and LG-SSM, are simultaneously updated by maximizing the evidence lower bound of the marginal likelihood. In contrast to the original model, we explain the motion with a spatial transformation from one image to another. This results in sharper reconstructions and the possibility of transferring auxiliary information, such as segmentation, through the image sequence. Our experiments, on cardiac ultrasound time series, show that the dynamic model outperforms traditional image registration in execution time, to a similar performance. Further, our model offers the possibility to impute and extrapolate for missing samples., Comment: published in 2022 25th International Conference on Information Fusion (FUSION)

The performance of neural network models is often limited by the availability of big data sets. To treat this problem, we survey and develop novel synthetic data generation and augmentation techniques for enhancing low/zero-sample learning in satellite imagery. In addition to extending synthetic data generation approaches, we propose a hierarchical detection approach to improve the utility of synthetic training samples. We consider existing techniques for producing synthetic imagery--3D models and neural style transfer--as well as introducing our own adversarially trained reskinning network, the GAN-Reskinner, to blend 3D models. Additionally, we test the value of synthetic data in a two-stage, hierarchical detection/classification model of our own construction. To test the effectiveness of synthetic imagery, we employ it in the training of detection models and our two stage model, and evaluate the resulting models on real satellite images. All modalities of synthetic data are tested extensively on practical, geospatial analysis problems. Our experiments show that synthetic data developed using our approach can often enhance detection performance, particularly when combined with some real training images. When the only source of data is synthetic, our GAN-Reskinner often boosts performance over conventionally rendered 3D models and in all cases the hierarchical model outperforms the baseline end-to-end detection architecture., Comment: 10 pages, 6 figures

Accurate 3D object detection (3DOD) is crucial for safe navigation of complex environments by autonomous robots. Regressing accurate 3D bounding boxes in cluttered environments based on sparse LiDAR data is however a highly challenging problem. We address this task by exploring recent advances in conditional energy-based models (EBMs) for probabilistic regression. While methods employing EBMs for regression have demonstrated impressive performance on 2D object detection in images, these techniques are not directly applicable to 3D bounding boxes. In this work, we therefore design a differentiable pooling operator for 3D bounding boxes, serving as the core module of our EBM network. We further integrate this general approach into the state-of-the-art 3D object detector SA-SSD. On the KITTI dataset, our proposed approach consistently outperforms the SA-SSD baseline across all 3DOD metrics, demonstrating the potential of EBM-based regression for highly accurate 3DOD. Code is available at https://github.com/fregu856/ebms_3dod., Comment: CVPR Workshops 2021. Code is available at https://github.com/fregu856/ebms_3dod

Energy-based models (EBMs) have become increasingly popular within computer vision in recent years. While they are commonly employed for generative image modeling, recent work has applied EBMs also for regression tasks, achieving state-of-the-art performance on object detection and visual tracking. Training EBMs is however known to be challenging. While a variety of different techniques have been explored for generative modeling, the application of EBMs to regression is not a well-studied problem. How EBMs should be trained for best possible regression performance is thus currently unclear. We therefore accept the task of providing the first detailed study of this problem. To that end, we propose a simple yet highly effective extension of noise contrastive estimation, and carefully compare its performance to six popular methods from literature on the tasks of 1D regression and object detection. The results of this comparison suggest that our training method should be considered the go-to approach. We also apply our method to the visual tracking task, achieving state-of-the-art performance on five datasets. Notably, our tracker achieves 63.7% AUC on LaSOT and 78.7% Success on TrackingNet. Code is available at https://github.com/fregu856/ebms_regression., Comment: BMVC 2020. Code is available at https://github.com/fregu856/ebms_regression

Our anatomy is in constant motion. With modern MR imaging it is possible to record this motion in real-time during an ongoing radiation therapy session. In this paper we present an image registration method that exploits the sequential nature of 2D MR images to estimate the corresponding displacement field. The method employs several discriminative correlation filters that independently track specific points. Together with a sparse-to-dense interpolation scheme we can then estimate of the displacement field. The discriminative correlation filters are trained online, and our method is modality agnostic. For the interpolation scheme we use a neural network with normalized convolutions that is trained using synthetic diffeomorphic displacement fields. The method is evaluated on a segmented cardiac dataset and when compared to two conventional methods we observe an improved performance. This improvement is especially pronounced when it comes to the detection of larger motions of small objects., Comment: Currently under review for a conference

While deep learning-based classification is generally tackled using standardized approaches, a wide variety of techniques are employed for regression. In computer vision, one particularly popular such technique is that of confidence-based regression, which entails predicting a confidence value for each input-target pair (x,y). While this approach has demonstrated impressive results, it requires important task-dependent design choices, and the predicted confidences lack a natural probabilistic meaning. We address these issues by proposing a general and conceptually simple regression method with a clear probabilistic interpretation. In our proposed approach, we create an energy-based model of the conditional target density p(y|x), using a deep neural network to predict the un-normalized density from (x,y). This model of p(y|x) is trained by directly minimizing the associated negative log-likelihood, approximated using Monte Carlo sampling. We perform comprehensive experiments on four computer vision regression tasks. Our approach outperforms direct regression, as well as other probabilistic and confidence-based methods. Notably, our model achieves a 2.2% AP improvement over Faster-RCNN for object detection on the COCO dataset, and sets a new state-of-the-art on visual tracking when applied for bounding box estimation. In contrast to confidence-based methods, our approach is also shown to be directly applicable to more general tasks such as age and head-pose estimation. Code is available at https://github.com/fregu856/ebms_regression., Comment: ECCV 2020. Code is available at https://github.com/fregu856/ebms_regression

While deep neural networks have become the go-to approach in computer vision, the vast majority of these models fail to properly capture the uncertainty inherent in their predictions. Estimating this predictive uncertainty can be crucial, for example in automotive applications. In Bayesian deep learning, predictive uncertainty is commonly decomposed into the distinct types of aleatoric and epistemic uncertainty. The former can be estimated by letting a neural network output the parameters of a certain probability distribution. Epistemic uncertainty estimation is a more challenging problem, and while different scalable methods recently have emerged, no extensive comparison has been performed in a real-world setting. We therefore accept this task and propose a comprehensive evaluation framework for scalable epistemic uncertainty estimation methods in deep learning. Our proposed framework is specifically designed to test the robustness required in real-world computer vision applications. We also apply this framework to provide the first properly extensive and conclusive comparison of the two current state-of-the-art scalable methods: ensembling and MC-dropout. Our comparison demonstrates that ensembling consistently provides more reliable and practically useful uncertainty estimates. Code is available at https://github.com/fregu856/evaluating_bdl., Comment: CVPR Workshops 2020. Code is available at https://github.com/fregu856/evaluating_bdl

In this paper, we present a framework to control a self-driving car by fusing raw information from RGB images and depth maps. A deep neural network architecture is used for mapping the vision and depth information, respectively, to steering commands. This fusion of information from two sensor sources allows to provide redundancy and fault tolerance in the presence of sensor failures. Even if one of the input sensors fails to produce the correct output, the other functioning sensor would still be able to maneuver the car. Such redundancy is crucial in the critical application of self-driving cars. The experimental results have showed that our method is capable of learning to use the relevant sensor information even when one of the sensors fail without any explicit signal.

Semantic segmentation maps can be used as input to models for maneuvering the controls of a car. However, not all labels may be necessary for making the control decision. One would expect that certain labels such as road lanes or sidewalks would be more critical in comparison with labels for vegetation or buildings which may not have a direct influence on the car's driving decision. In this appendix, we evaluate and quantify how sensitive and important the different semantic labels are for controlling the car. Labels that do not influence the driving decision are remapped to other classes, thereby simplifying the task by reducing to only labels critical for driving of the vehicle.

Model-free reinforcement learning has recently been shown to successfully learn navigation policies from raw sensor data. In this work, we address the problem of learning driving policies for an autonomous agent in a high-fidelity simulator. Building upon recent research that applies deep reinforcement learning to navigation problems, we present a modular deep reinforcement learning approach to predict the steering angle of the car from raw images. The first module extracts a low-dimensional latent semantic representation of the image. The control module trained with reinforcement learning takes the latent vector as input to predict the correct steering angle. The experimental results have showed that our method is capable of learning to maneuver the car without any human control signals.

In this work, we consider the inverse problem of reconstructing the internal structure of an object from limited x-ray projections. We use a Gaussian process prior to model the target function and estimate its (hyper)parameters from measured data. In contrast to other established methods, this comes with the advantage of not requiring any manual parameter tuning, which usually arises in classical regularization strategies. Our method uses a basis function expansion technique for the Gaussian process which significantly reduces the computational complexity and avoids the need for numerical integration. The approach also allows for reformulation of come classical regularization methods as Laplacian and Tikhonov regularization as Gaussian process regression, and hence provides an efficient algorithm and principled means for their parameter tuning. Results from simulated and real data indicate that this approach is less sensitive to streak artifacts as compared to the commonly used method of filtered backprojection.

We present a method that gets as input an audio of violin or piano playing, and outputs a video of skeleton predictions which are further used to animate an avatar. The key idea is to create an animation of an avatar that moves their hands similarly to how a pianist or violinist would do, just from audio. Aiming for a fully detailed correct arms and fingers motion is a goal, however, it's not clear if body movement can be predicted from music at all. In this paper, we present the first result that shows that natural body dynamics can be predicted at all. We built an LSTM network that is trained on violin and piano recital videos uploaded to the Internet. The predicted points are applied onto a rigged avatar to create the animation., Comment: Link with videos https://arviolin.github.io/AudioBodyDynamics/

Data-efficient reinforcement learning (RL) in continuous state-action spaces using very high-dimensional observations remains a key challenge in developing fully autonomous systems. We consider a particularly important instance of this challenge, the pixels-to-torques problem, where an RL agent learns a closed-loop control policy ("torques") from pixel information only. We introduce a data-efficient, model-based reinforcement learning algorithm that learns such a closed-loop policy directly from pixel information. The key ingredient is a deep dynamical model for learning a low-dimensional feature embedding of images jointly with a predictive model in this low-dimensional feature space. Joint learning is crucial for long-term predictions, which lie at the core of the adaptive nonlinear model predictive control strategy that we use for closed-loop control. Compared to state-of-the-art RL methods for continuous states and actions, our approach learns quickly, scales to high-dimensional state spaces, is lightweight and an important step toward fully autonomous end-to-end learning from pixels to torques.