Tom Burgert

Tom Burgert, Mahdyar Ravanbakhsh, Begüm Demir

The development of accurate methods for multi-label classification (MLC) of remote sensing (RS) images is one of the most important research topics in RS. To address MLC problems, the use of deep neural networks that require a high number of reliable training images annotated by multiple land-cover class labels (multi-labels) has been found popular in RS. However, collecting such annotations is time-consuming and costly. A common procedure to obtain annotations at zero labeling cost is to rely on thematic products or crowdsourced labels. As a drawback, these procedures come with the risk of label noise that can distort the learning process of the MLC algorithms. In the literature, most label noise robust methods are designed for single-label classification (SLC) problems in computer vision (CV), where each image is annotated by a single label. Unlike SLC, label noise in MLC can be associated with: 1) subtractive label-noise (a land cover class label is not assigned to an image while that class is present in the image); 2) additive label-noise (a land cover class label is assigned to an image although that class is not present in the given image); and 3) mixed label-noise (a combination of both). In this paper, we investigate three different noise robust CV SLC methods and adapt them to be robust for multi-label noise scenarios in RS. During experiments, we study the effects of different types of multi-label noise and evaluate the adapted methods rigorously. To this end, we also introduce a synthetic multi-label noise injection strategy that is more adequate to simulate operational scenarios compared to the uniform label noise injection strategy, in which the labels of absent and present classes are flipped at uniform probability. Further, we study the relevance of different evaluation metrics in MLC problems under noisy multi-labels.

Kai Norman Clasen, Leonard Hackel, Tom Burgert et al.

This paper presents refined BigEarthNet (reBEN) that is a large-scale, multi-modal remote sensing dataset constructed to support deep learning (DL) studies for remote sensing image analysis. The reBEN dataset consists of 549,488 pairs of Sentinel-1 and Sentinel-2 image patches. To construct reBEN, we initially consider the Sentinel-1 and Sentinel-2 tiles used to construct the BigEarthNet dataset and then divide them into patches of size 1200 m x 1200 m. We apply atmospheric correction to the Sentinel-2 patches using the latest version of the sen2cor tool, resulting in higher-quality patches compared to those present in BigEarthNet. Each patch is then associated with a pixel-level reference map and scene-level multi-labels. This makes reBEN suitable for pixel- and scene-based learning tasks. The labels are derived from the most recent CORINE Land Cover (CLC) map of 2018 by utilizing the 19-class nomenclature as in BigEarthNet. The use of the most recent CLC map results in overcoming the label noise present in BigEarthNet. Furthermore, we introduce a new geographical-based split assignment algorithm that significantly reduces the spatial correlation among the train, validation, and test sets with respect to those present in BigEarthNet. This increases the reliability of the evaluation of DL models. To minimize the DL model training time, we introduce software tools that convert the reBEN dataset into a DL-optimized data format. In our experiments, we show the potential of reBEN for multi-modal multi-label image classification problems by considering several state-of-the-art DL models. The pre-trained model weights, associated code, and complete dataset are available at https://bigearth.net.

Tom Burgert, Leonard Hackel, Paolo Rota et al.

Self-supervised learning (SSL) has become a powerful paradigm for learning from large, unlabeled datasets, particularly in computer vision (CV). However, applying SSL to multispectral remote sensing (RS) images presents unique challenges and opportunities due to the geographical and temporal variability of the data. In this paper, we introduce GeoRank, a novel regularization method for contrastive SSL that improves upon prior techniques by directly optimizing spherical distances to embed geographical relationships into the learned feature space. GeoRank outperforms or matches prior methods that integrate geographical metadata and consistently improves diverse contrastive SSL algorithms (e.g., BYOL, DINO). Beyond this, we present a systematic investigation of key adaptations of contrastive SSL for multispectral RS images, including the effectiveness of data augmentations, the impact of dataset cardinality and image size on performance, and the task dependency of temporal views. Code is available at https://github.com/tomburgert/georank.

Tom Burgert, Kai Norman Clasen, Jonas Klotz et al.

The development of supervised deep learning-based methods for multi-label scene classification (MLC) is one of the prominent research directions in remote sensing (RS). However, collecting annotations for large RS image archives is time-consuming and costly. To address this issue, several data augmentation methods have been introduced in RS. Among others, the CutMix data augmentation technique, which combines parts of two existing training images to generate an augmented image, stands out as a particularly effective approach. However, the direct application of CutMix in RS MLC can lead to the erasure or addition of class labels (i.e., label noise) in the augmented (i.e., combined) training image. To address this problem, we introduce a label propagation (LP) strategy that allows the effective application of CutMix in the context of MLC problems in RS without being affected by label noise. To this end, our proposed LP strategy exploits pixel-level class positional information to update the multi-label of the augmented training image. We propose to access such class positional information from reference maps (e.g., thematic products) associated with each training image or from class explanation masks provided by an explanation method if no reference maps are available. Similarly to pairing two training images, our LP strategy carries out a pairing operation on the associated pixel-level class positional information to derive the updated multi-label for the augmented image. Experimental results show the effectiveness of our LP strategy in general (e.g., an improvement of 2% to 4% mAP macro compared to standard CutMix) and its robustness in the case of various simulated and real scenarios with noisy class positional information in particular. Code is available at https://git.tu-berlin.de/rsim/cutmix_lp.

Tom Burgert, Julia Henkel, Begüm Demir

The development of reliable methods for multi-label classification (MLC) has become a prominent research direction in remote sensing (RS). As the scale of RS data continues to expand, annotation procedures increasingly rely on thematic products or crowdsourced procedures to reduce the cost of manual annotation. While cost-effective, these strategies often introduce multi-label noise in the form of partially incorrect annotations. In MLC, label noise arises as additive noise, subtractive noise, or a combination of both in the form of mixed noise. Previous work has largely overlooked this distinction and commonly treats noisy annotations as supervised signals, lacking mechanisms that explicitly adapt learning behavior to different noise types. To address this limitation, we propose NAR, a noise-adaptive regularization method that explicitly distinguishes between additive and subtractive noise within a semi-supervised learning framework. NAR employs a confidence-based label handling mechanism that dynamically retains label entries with high confidence, temporarily deactivates entries with moderate confidence, and corrects low confidence entries via flipping. This selective attenuation of supervision is integrated with early-learning regularization (ELR) to stabilize training and mitigate overfitting to corrupted labels. Experiments across additive, subtractive, and mixed noise scenarios demonstrate that NAR consistently improves robustness compared with existing methods. Performance improvements are most pronounced under subtractive and mixed noise, indicating that adaptive suppression and selective correction of noisy supervision provide an effective strategy for noise robust learning in RS MLC.

Leonard Hackel, Tom Burgert, Begüm Demir

Large-scale foundation models (FMs) in remote sensing (RS) are developed based on the paradigms established in computer vision (CV) and have shown promise for various Earth observation applications. However, the direct transfer of scaling assumptions from CV to RS has not been adequately examined. We hypothesize that RS FMs enter an overparameterized regime at substantially smaller scales than their CV counterparts, where increasing parameter count primarily induces redundant representations rather than qualitatively new abstractions. To test this hypothesis, we use post-hoc slimming, where we uniformly reduce the width of pretrained encoder, as a tool to measure representational redundancy across six state-of-the-art RS FMs on four downstream classification tasks. Our findings reveal a significant contrast with those in the CV domain: while a post-hoc slimmed masked autoencoder (MAE) trained on ImageNet retains less than 10% accuracy at 1% FLOPs, RS FMs maintain over 71% relative accuracy at the same budget. This sevenfold difference provides strong empirical support for our hypothesis. We further demonstrate that learned slimmable training can improve both Momentum Contrast (MoCo)- and MAE- based models. In addition, through the explained variance ratio and the feature correlation analysis, we provide mechanistic explanations showing that RS FMs distribute task-relevant information with high redundancy. Our findings establish post-hoc slimmability as both a practical deployment strategy for resource-constrained environments and a diagnostic tool that challenges the prevailing scaling paradigm in RS. Upon acceptance, we will publish all code.

Tom Burgert, Oliver Stoll, Paolo Rota et al.

The hypothesis that Convolutional Neural Networks (CNNs) are inherently texture-biased has shaped much of the discourse on feature use in deep learning. We revisit this hypothesis by examining limitations in the cue-conflict experiment by Geirhos et al. To address these limitations, we propose a domain-agnostic framework that quantifies feature reliance through systematic suppression of shape, texture, and color cues, avoiding the confounds of forced-choice conflicts. By evaluating humans and neural networks under controlled suppression conditions, we find that CNNs are not inherently texture-biased but predominantly rely on local shape features. Nonetheless, this reliance can be substantially mitigated through modern training strategies or architectures (ConvNeXt, ViTs). We further extend the analysis across computer vision, medical imaging, and remote sensing, revealing that reliance patterns differ systematically: computer vision models prioritize shape, medical imaging models emphasize color, and remote sensing models exhibit a stronger reliance on texture. Code is available at https://github.com/tomburgert/feature-reliance.

Tom Burgert, Begüm Demir

The application of data augmentation for deep learning (DL) methods plays an important role in achieving state-of-the-art results in supervised, semi-supervised, and self-supervised image classification. In particular, channel transformations (e.g., solarize, grayscale, brightness adjustments) are integrated into data augmentation pipelines for remote sensing (RS) image classification tasks. However, contradicting beliefs exist about their proper applications to RS images. A common point of critique is that the application of channel augmentation techniques may lead to physically inconsistent spectral data (i.e., pixel signatures). To shed light on the open debate, we propose an approach to estimate whether a channel augmentation technique affects the physical information of RS images. To this end, the proposed approach estimates a score that measures the alignment of a pixel signature within a time series that can be naturally subject to deviations caused by factors such as acquisition conditions or phenological states of vegetation. We compare the scores associated with original and augmented pixel signatures to evaluate the physical consistency. Experimental results on a multi-label image classification task show that channel augmentations yielding a score that exceeds the expected deviation of original pixel signatures can not improve the performance of a baseline model trained without augmentation.

Leonard Hackel, Tom Burgert, Begüm Demir

Self-supervised learning through masked autoencoders has attracted great attention for remote sensing (RS) foundation model (FM) development, enabling improved representation learning across diverse sensors and downstream tasks. However, existing RS FMs often either suffer from substantial computational complexity during both training and inference or exhibit limited representational capacity. These issues restrict their practical applicability in RS. To address this limitation, we propose an adaptation for enhancing the efficiency of RS FMs by integrating the Soft mixture-of-experts (MoE) mechanism into the FM. The integration of Soft MoEs into the FM allows modality-specific expert specialization alongside shared cross-sensor representation learning. To demonstrate the effectiveness of our adaptation, we apply it on the Cross-Sensor Masked Autoencoder (CSMAE) model, resulting in the Cross-Sensor Mixture-of-Experts (CSMoE) model. In addition, we introduce a thematic-climatic descriptor-driven sampling strategy for the construction of a representative and diverse training set to train our CSMoE model. Extensive experiments on scene classification, semantic segmentation, and content-based image retrieval demonstrate that our adaptation yields a reduction in computational requirements while maintaining or improving representational performance. Compared to state-of-the-art RS FMs, CSMoE achieves a superior trade-off between representational capacity, accuracy, and computational efficiency. On average, CSMoE achieves more than twice the computational efficiency of existing RS FMs, while maintaining competitive performance across all experiments. These results show the effectiveness of the proposed adaptation for creating computationally efficient RS FMs. The code for the model, the training set creation, and the model weights will be available at https://git.tu-berlin.de/rsim/csmoe.

Jonas Klotz, Tom Burgert, Begüm Demir

The development of explainable artificial intelligence (xAI) methods for scene classification problems has attracted great attention in remote sensing (RS). Most xAI methods and the related evaluation metrics in RS are initially developed for natural images considered in computer vision (CV), and their direct usage in RS may not be suitable. To address this issue, in this paper, we investigate the effectiveness of explanation methods and metrics in the context of RS image scene classification. In detail, we methodologically and experimentally analyze ten explanation metrics spanning five categories (faithfulness, robustness, localization, complexity, randomization), applied to five established feature attribution methods (Occlusion, LIME, GradCAM, LRP, and DeepLIFT) across three RS datasets. Our methodological analysis identifies key limitations in both explanation methods and metrics. The performance of perturbation-based methods, such as Occlusion and LIME, heavily depends on perturbation baselines and spatial characteristics of RS scenes. Gradient-based approaches like GradCAM struggle when multiple labels are present in the same image, while some relevance propagation methods (LRP) can distribute relevance disproportionately relative to the spatial extent of classes. Analogously, we find limitations in evaluation metrics. Faithfulness metrics share the same problems as perturbation-based methods. Localization metrics and complexity metrics are unreliable for classes with a large spatial extent. In contrast, robustness metrics and randomization metrics consistently exhibit greater stability. Our experimental results support these methodological findings. Based on our analysis, we provide guidelines for selecting explanation methods, metrics, and hyperparameters in the context of RS image scene classification.