Search

We propose an interactive approach for 3D instance segmentation, where users can iteratively collaborate with a deep learning model to segment objects in a 3D point cloud directly. Current methods for 3D instance segmentation are generally trained in a fully-supervised fashion, which requires large amounts of costly training labels, and does not generalize well to classes unseen during training. Few works have attempted to obtain 3D segmentation masks using human interactions. Existing methods rely on user feedback in the 2D image domain. As a consequence, users are required to constantly switch between 2D images and 3D representations, and custom architectures are employed to combine multiple input modalities. Therefore, integration with existing standard 3D models is not straightforward. The core idea of this work is to enable users to interact directly with 3D point clouds by clicking on desired 3D objects of interest~(or their background) to interactively segment the scene in an open-world setting. Specifically, our method does not require training data from any target domain, and can adapt to new environments where no appropriate training sets are available. Our system continuously adjusts the object segmentation based on the user feedback and achieves accurate dense 3D segmentation masks with minimal human effort (few clicks per object). Besides its potential for efficient labeling of large-scale and varied 3D datasets, our approach, where the user directly interacts with the 3D environment, enables new applications in AR/VR and human-robot interaction., Comment: 2023 IEEE International Conference on Robotics and Automation (ICRA)

The worldwide variation in vegetation height is fundamental to the global carbon cycle and central to the functioning of ecosystems and their biodiversity. Geospatially explicit and, ideally, highly resolved information is required to manage terrestrial ecosystems, mitigate climate change, and prevent biodiversity loss. Here, we present the first global, wall-to-wall canopy height map at 10 m ground sampling distance for the year 2020. No single data source meets these requirements: dedicated space missions like GEDI deliver sparse height data, with unprecedented coverage, whereas optical satellite images like Sentinel-2 offer dense observations globally, but cannot directly measure vertical structures. By fusing GEDI with Sentinel-2, we have developed a probabilistic deep learning model to retrieve canopy height from Sentinel-2 images anywhere on Earth, and to quantify the uncertainty in these estimates. The presented approach reduces the saturation effect commonly encountered when estimating canopy height from satellite images, allowing to resolve tall canopies with likely high carbon stocks. According to our map, only 5% of the global landmass is covered by trees taller than 30 m. Such data play an important role for conservation, e.g., we find that only 34% of these tall canopies are located within protected areas. Our model enables consistent, uncertainty-informed worldwide mapping and supports an ongoing monitoring to detect change and inform decision making. The approach can serve ongoing efforts in forest conservation, and has the potential to foster advances in climate, carbon, and biodiversity modelling.

Our goal is to recover the 3D shape and pose of dogs from a single image. This is a challenging task because dogs exhibit a wide range of shapes and appearances, and are highly articulated. Recent work has proposed to directly regress the SMAL animal model, with additional limb scale parameters, from images. Our method, called BARC (Breed-Augmented Regression using Classification), goes beyond prior work in several important ways. First, we modify the SMAL shape space to be more appropriate for representing dog shape. But, even with a better shape model, the problem of regressing dog shape from an image is still challenging because we lack paired images with 3D ground truth. To compensate for the lack of paired data, we formulate novel losses that exploit information about dog breeds. In particular, we exploit the fact that dogs of the same breed have similar body shapes. We formulate a novel breed similarity loss consisting of two parts: One term encourages the shape of dogs from the same breed to be more similar than dogs of different breeds. The second one, a breed classification loss, helps to produce recognizable breed-specific shapes. Through ablation studies, we find that our breed losses significantly improve shape accuracy over a baseline without them. We also compare BARC qualitatively to WLDO with a perceptual study and find that our approach produces dogs that are significantly more realistic. This work shows that a-priori information about genetic similarity can help to compensate for the lack of 3D training data. This concept may be applicable to other animal species or groups of species. Our code is publicly available for research purposes at https://barc.is.tue.mpg.de/., Comment: accepted for publication at CVPR 2022

We introduce a novel formulation for guided super-resolution. Its core is a differentiable optimisation layer that operates on a learned affinity graph. The learned graph potentials make it possible to leverage rich contextual information from the guide image, while the explicit graph optimisation within the architecture guarantees rigorous fidelity of the high-resolution target to the low-resolution source. With the decision to employ the source as a constraint rather than only as an input to the prediction, our method differs from state-of-the-art deep architectures for guided super-resolution, which produce targets that, when downsampled, will only approximately reproduce the source. This is not only theoretically appealing, but also produces crisper, more natural-looking images. A key property of our method is that, although the graph connectivity is restricted to the pixel lattice, the associated edge potentials are learned with a deep feature extractor and can encode rich context information over large receptive fields. By taking advantage of the sparse graph connectivity, it becomes possible to propagate gradients through the optimisation layer and learn the edge potentials from data. We extensively evaluate our method on several datasets, and consistently outperform recent baselines in terms of quantitative reconstruction errors, while also delivering visually sharper outputs. Moreover, we demonstrate that our method generalises particularly well to new datasets not seen during training., Comment: CVPR 2022

High-resolution optical satellite sensors, combined with dense stereo algorithms, have made it possible to reconstruct 3D city models from space. However, these models are, in practice, rather noisy and tend to miss small geometric features that are clearly visible in the images. We argue that one reason for the limited quality may be a too early, heuristic reduction of the triangulated 3D point cloud to an explicit height field or surface mesh. To make full use of the point cloud and the underlying images, we introduce ImpliCity, a neural representation of the 3D scene as an implicit, continuous occupancy field, driven by learned embeddings of the point cloud and a stereo pair of ortho-photos. We show that this representation enables the extraction of high-quality DSMs: with image resolution 0.5$\,$m, ImpliCity reaches a median height error of $\approx\,$0.7$\,$m and outperforms competing methods, especially w.r.t. building reconstruction, featuring intricate roof details, smooth surfaces, and straight, regular outlines., Comment: Accepted for publication in the International Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences (camera-ready version including keywords + supplementary material)

Monitoring and managing Earth's forests in an informed manner is an important requirement for addressing challenges like biodiversity loss and climate change. While traditional in situ or aerial campaigns for forest assessments provide accurate data for analysis at regional level, scaling them to entire countries and beyond with high temporal resolution is hardly possible. In this work, we propose a method based on deep ensembles that densely estimates forest structure variables at country-scale with 10-meter resolution, using freely available satellite imagery as input. Our method jointly transforms Sentinel-2 optical images and Sentinel-1 synthetic-aperture radar images into maps of five different forest structure variables: 95th height percentile, mean height, density, Gini coefficient, and fractional cover. We train and test our model on reference data from 41 airborne laser scanning missions across Norway and demonstrate that it is able to generalize to unseen test regions, achieving normalized mean absolute errors between 11% and 15%, depending on the variable. Our work is also the first to propose a variant of so-called Bayesian deep learning to densely predict multiple forest structure variables with well-calibrated uncertainty estimates from satellite imagery. The uncertainty information increases the trustworthiness of the model and its suitability for downstream tasks that require reliable confidence estimates as a basis for decision making. We present an extensive set of experiments to validate the accuracy of the predicted maps as well as the quality of the predicted uncertainties. To demonstrate scalability, we provide Norway-wide maps for the five forest structure variables.

We investigate the possibility of 3D scene reconstruction from two or more overlapping webcam streams. A large, and growing, number of webcams observe places of interest and are publicly accessible. The question naturally arises: can we make use of this free data source for 3D computer vision? It turns out that the task to reconstruct scene structure from webcam streams is very different from standard structure-from-motion (SfM), and conventional SfM pipelines fail. In the webcam setting there are very few views of the same scene, in most cases only the minimum of two. These viewpoints often have large baselines and/or scale differences, their overlap is rather limited, and besides unknown internal and external calibration also their temporal synchronisation is unknown. On the other hand, they record rather large fields of view continuously over long time spans, so that they regularly observe dynamic objects moving through the scene. We show how to leverage recent advances in several areas of computer vision to adapt SfM reconstruction to this particular scenario and reconstruct the unknown camera poses, the 3D scene structure, and the 3D trajectories of dynamic objects.

Humanitarian actions require accurate information to efficiently delegate support operations. Such information can be maps of building footprints, building functions, and population densities. While the access to this information is comparably easy in industrialized countries thanks to reliable census data and national geo-data infrastructures, this is not the case for developing countries, where that data is often incomplete or outdated. Building maps derived from remote sensing images may partially remedy this challenge in such countries, but are not always accurate due to different landscape configurations and lack of validation data. Even when they exist, building footprint layers usually do not reveal more fine-grained building properties, such as the number of stories or the building's function (e.g., office, residential, school, etc.). In this project we aim to automate building footprint and function mapping using heterogeneous data sources. In a first step, we intend to delineate buildings from satellite data, using deep learning models for semantic image segmentation. Building functions shall be retrieved by parsing social media data like for instance tweets, as well as ground-based imagery, to automatically identify different buildings functions and retrieve further information such as the number of building stories. Building maps augmented with those additional attributes make it possible to derive more accurate population density maps, needed to support the targeted provision of humanitarian aid.

Fusing satellite imagery acquired with different sensors has been a long-standing challenge of Earth observation, particularly across different modalities such as optical and Synthetic Aperture Radar (SAR) images. Here, we explore the joint analysis of imagery from different sensors in the light of representation learning: we propose to learn a joint embedding of multiple satellite sensors within a deep neural network. Our application problem is the monitoring of lake ice on Alpine lakes. To reach the temporal resolution requirement of the Swiss Global Climate Observing System (GCOS) office, we combine three image sources: Sentinel-1 SAR (S1-SAR), Terra MODIS, and Suomi-NPP VIIRS. The large gaps between the optical and SAR domains and between the sensor resolutions make this a challenging instance of the sensor fusion problem. Our approach can be classified as a late fusion that is learned in a data-driven manner. The proposed network architecture has separate encoding branches for each image sensor, which feed into a single latent embedding. I.e., a common feature representation shared by all inputs, such that subsequent processing steps deliver comparable output irrespective of which sort of input image was used. By fusing satellite data, we map lake ice at a temporal resolution of < 1.5 days. The network produces spatially explicit lake ice maps with pixel-wise accuracies > 91% (respectively, mIoU scores > 60%) and generalises well across different lakes and winters. Moreover, it sets a new state-of-the-art for determining the important ice-on and ice-off dates for the target lakes, in many cases meeting the GCOS requirement.

The increasing demand for commodities is leading to changes in land use worldwide. In the tropics, deforestation, which causes high carbon emissions and threatens biodiversity, is often linked to agricultural expansion. While the need for deforestation-free global supply chains is widely recognized, making progress in practice remains a challenge. Here, we propose an automated approach that aims to support conservation and sustainable land use planning decisions by mapping tropical landscapes at large scale and high spatial resolution following the High Carbon Stock (HCS) approach. A deep learning approach is developed that estimates canopy height for each 10 m Sentinel-2 pixel by learning from sparse GEDI LIDAR reference data, achieving an overall RMSE of 6.3 m. We show that these wall-to-wall maps of canopy top height are predictive for classifying HCS forests and degraded areas with an overall accuracy of 86 % and produce a first high carbon stock map for Indonesia, Malaysia, and the Philippines.

Modern optical satellite sensors enable high-resolution stereo reconstruction from space. But the challenging imaging conditions when observing the Earth from space push stereo matching to its limits. In practice, the resulting digital surface models (DSMs) are fairly noisy and often do not attain the accuracy needed for high-resolution applications such as 3D city modeling. Arguably, stereo correspondence based on low-level image similarity is insufficient and should be complemented with a-priori knowledge about the expected surface geometry beyond basic local smoothness. To that end, we introduce ResDepth, a convolutional neural network that learns such an expressive geometric prior from example data. ResDepth refines an initial, raw stereo DSM while conditioning the refinement on the images. I.e., it acts as a smart, learned post-processing filter and can seamlessly complement any stereo matching pipeline. In a series of experiments, we find that the proposed method consistently improves stereo DSMs both quantitatively and qualitatively. We show that the prior encoded in the network weights captures meaningful geometric characteristics of urban design, which also generalize across different districts and even from one city to another. Moreover, we demonstrate that, by training on a variety of stereo pairs, ResDepth can acquire a sufficient degree of invariance against variations in imaging conditions and acquisition geometry., Comment: Accepted for publication in the ISPRS Journal of Photogrammetry and Remote Sensing

Automatic identification of plant specimens from amateur photographs could improve species range maps, thus supporting ecosystems research as well as conservation efforts. However, classifying plant specimens based on image data alone is challenging: some species exhibit large variations in visual appearance, while at the same time different species are often visually similar; additionally, species observations follow a highly imbalanced, long-tailed distribution due to differences in abundance as well as observer biases. On the other hand, most species observations are accompanied by side information about the spatial, temporal and ecological context. Moreover, biological species are not an unordered list of classes but embedded in a hierarchical taxonomic structure. We propose a multimodal deep learning model that takes into account these additional cues in a unified framework. Our Digital Taxonomist is able to identify plant species in photographs better than a classifier trained on the image content alone, the performance gained is over 6 percent points in terms of accuracy., Comment: Accepted for publication in the ISPRS Journal of Photogrammetry and Remote Sensing

We propose an end-to-end trainable framework that processes large-scale visual data tensors by looking at a fraction of their entries only. Our method combines a neural network encoder with a tensor train decomposition to learn a low-rank latent encoding, coupled with cross-approximation (CA) to learn the representation through a subset of the original samples. CA is an adaptive sampling algorithm that is native to tensor decompositions and avoids working with the full high-resolution data explicitly. Instead, it actively selects local representative samples that we fetch out-of-core and on-demand. The required number of samples grows only logarithmically with the size of the input. Our implicit representation of the tensor in the network enables processing large grids that could not be otherwise tractable in their uncompressed form. The proposed approach is particularly useful for large-scale multidimensional grid data (e.g., 3D tomography), and for tasks that require context over a large receptive field (e.g., predicting the medical condition of entire organs). The code is available at https://github.com/aelphy/c-pic.

Accurate mapping of oil palm is important for understanding its past and future impact on the environment. We propose to map and count oil palms by estimating tree densities per pixel for large-scale analysis. This allows for fine-grained analysis, for example regarding different planting patterns. To that end, we propose a new, active deep learning method to estimate oil palm density at large scale from Sentinel-2 satellite images, and apply it to generate complete maps for Malaysia and Indonesia. What makes the regression of oil palm density challenging is the need for representative reference data that covers all relevant geographical conditions across a large territory. Specifically for density estimation, generating reference data involves counting individual trees. To keep the associated labelling effort low we propose an active learning (AL) approach that automatically chooses the most relevant samples to be labelled. Our method relies on estimates of the epistemic model uncertainty and of the diversity among samples, making it possible to retrieve an entire batch of relevant samples in a single iteration. Moreover, our algorithm has linear computational complexity and is easily parallelisable to cover large areas. We use our method to compute the first oil palm density map with $10\,$m Ground Sampling Distance (GSD) , for all of Indonesia and Malaysia and for two different years, 2017 and 2019. The maps have a mean absolute error of $\pm$7.3 trees/$ha$, estimated from an independent validation set. We also analyse density variations between different states within a country and compare them to official estimates. According to our estimates there are, in total, $>1.2$ billion oil palms in Indonesia covering $>$15 million $ha$, and $>0.5$ billion oil palms in Malaysia covering $>6$ million $ha$.

Visual re-localization means using a single image as input to estimate the camera's location and orientation relative to a pre-recorded environment. The highest-scoring methods are "structure based," and need the query camera's intrinsics as an input to the model, with careful geometric optimization. When intrinsics are absent, methods vie for accuracy by making various other assumptions. This yields fairly good localization scores, but the models are "narrow" in some way, eg., requiring costly test-time computations, or depth sensors, or multiple query frames. In contrast, our proposed method makes few special assumptions, and is fairly lightweight in training and testing. Our pose regression network learns from only relative poses of training scenes. For inference, it builds a graph connecting the query image to training counterparts and uses a graph neural network (GNN) with image representations on nodes and image-pair representations on edges. By efficiently passing messages between them, both representation types are refined to produce a consistent camera pose estimate. We validate the effectiveness of our approach on both standard indoor (7-Scenes) and outdoor (Cambridge Landmarks) camera re-localization benchmarks. Our relative pose regression method matches the accuracy of absolute pose regression networks, while retaining the relative-pose models' test-time speed and ability to generalize to non-training scenes.

Depleting lake ice is a climate change indicator, just like sea-level rise or glacial retreat. Monitoring Lake Ice Phenology (LIP) is useful because long-term freezing and thawing patterns serve as sentinels to understand regional and global climate change. We report a study for the Oberengadin region of Switzerland, where several small- and medium-sized mountain lakes are located. We observe the LIP events, such as freeze-up, break-up and ice cover duration, across two decades (2000-2020) from optical satellite images. We analyse the time series of MODIS imagery by estimating spatially resolved maps of lake ice for these Alpine lakes with supervised machine learning. To train the classifier we rely on reference data annotated manually based on webcam images. From the ice maps, we derive long-term LIP trends. Since the webcam data are only available for two winters, we cross-check our results against the operational MODIS and VIIRS snow products. We find a change in complete freeze duration of -0.76 and -0.89 days per annum for lakes Sils and Silvaplana, respectively. Furthermore, we observe plausible correlations of the LIP trends with climate data measured at nearby meteorological stations. We notice that mean winter air temperature has a negative correlation with the freeze duration and break-up events and a positive correlation with the freeze-up events. Additionally, we observe a strong negative correlation of sunshine during the winter months with the freeze duration and break-up events., Comment: This version of the article has been accepted for publication, after peer review but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: http://dx.doi.org/10.1007/s41064-022-00215-x

Style transfer aims to render the content of a given image in the graphical/artistic style of another image. The fundamental concept underlying NeuralStyle Transfer (NST) is to interpret style as a distribution in the feature space of a Convolutional Neural Network, such that a desired style can be achieved by matching its feature distribution. We show that most current implementations of that concept have important theoretical and practical limitations, as they only partially align the feature distributions. We propose a novel approach that matches the distributions more precisely, thus reproducing the desired style more faithfully, while still being computationally efficient. Specifically, we adapt the dual form of Central Moment Discrepancy (CMD), as recently proposed for domain adaptation, to minimize the difference between the target style and the feature distribution of the output image. The dual interpretation of this metric explicitly matches all higher-order centralized moments and is therefore a natural extension of existing NST methods that only take into account the first and second moments. Our experiments confirm that the strong theoretical properties also translate to visually better style transfer, and better disentangle style from semantic image content.

NASA's Global Ecosystem Dynamics Investigation (GEDI) is a key climate mission whose goal is to advance our understanding of the role of forests in the global carbon cycle. While GEDI is the first space-based LIDAR explicitly optimized to measure vertical forest structure predictive of aboveground biomass, the accurate interpretation of this vast amount of waveform data across the broad range of observational and environmental conditions is challenging. Here, we present a novel supervised machine learning approach to interpret GEDI waveforms and regress canopy top height globally. We propose a probabilistic deep learning approach based on an ensemble of deep convolutional neural networks(CNN) to avoid the explicit modelling of unknown effects, such as atmospheric noise. The model learns to extract robust features that generalize to unseen geographical regions and, in addition, yields reliable estimates of predictive uncertainty. Ultimately, the global canopy top height estimates produced by our model have an expected RMSE of 2.7 m with low bias.

We introduce PC2WF, the first end-to-end trainable deep network architecture to convert a 3D point cloud into a wireframe model. The network takes as input an unordered set of 3D points sampled from the surface of some object, and outputs a wireframe of that object, i.e., a sparse set of corner points linked by line segments. Recovering the wireframe is a challenging task, where the numbers of both vertices and edges are different for every instance, and a-priori unknown. Our architecture gradually builds up the model: It starts by encoding the points into feature vectors. Based on those features, it identifies a pool of candidate vertices, then prunes those candidates to a final set of corner vertices and refines their locations. Next, the corners are linked with an exhaustive set of candidate edges, which is again pruned to obtain the final wireframe. All steps are trainable, and errors can be backpropagated through the entire sequence. We validate the proposed model on a publicly available synthetic dataset, for which the ground truth wireframes are accessible, as well as on a new real-world dataset. Our model produces wireframe abstractions of good quality and outperforms several baselines.

Recent years have seen a proliferation of new digital products for the efficient management of indoor spaces, with important applications like emergency management, virtual property showcasing and interior design. These products rely on accurate 3D models of the environments considered, including information on both architectural and non-permanent elements. These models must be created from measured data such as RGB-D images or 3D point clouds, whose capture and consolidation involves lengthy data workflows. This strongly limits the rate at which 3D models can be produced, preventing the adoption of many digital services for indoor space management. We provide an alternative to such data-intensive procedures by presenting Walk2Map, a data-driven approach to generate floor plans only from trajectories of a person walking inside the rooms. Thanks to recent advances in data-driven inertial odometry, such minimalistic input data can be acquired from the IMU readings of consumer-level smartphones, which allows for an effortless and scalable mapping of real-world indoor spaces. Our work is based on learning the latent relation between an indoor walk trajectory and the information represented in a floor plan: interior space footprint, portals, and furniture. We distinguish between recovering area-related (interior footprint, furniture) and wall-related (doors) information and use two different neural architectures for the two tasks: an image-based Encoder-Decoder and a Graph Convolutional Network, respectively. We train our networks using scanned 3D indoor models and apply them in a cascaded fashion on an indoor walk trajectory at inference time. We perform a qualitative and quantitative evaluation using both simulated and measured, real-world trajectories, and compare against a baseline method for image-to-image translation. The experiments confirm the feasibility of our approach., Comment: To be published in Computer Graphics Forum (Proc. Eurographics 2021)

The aim of this paper is to map agricultural crops by classifying satellite image time series. Domain experts in agriculture work with crop type labels that are organised in a hierarchical tree structure, where coarse classes (like orchards) are subdivided into finer ones (like apples, pears, vines, etc.). We develop a crop classification method that exploits this expert knowledge and significantly improves the mapping of rare crop types. The three-level label hierarchy is encoded in a convolutional, recurrent neural network (convRNN), such that for each pixel the model predicts three labels at different level of granularity. This end-to-end trainable, hierarchical network architecture allows the model to learn joint feature representations of rare classes (e.g., apples, pears) at a coarser level (e.g., orchard), thereby boosting classification performance at the fine-grained level. Additionally, labelling at different granularity also makes it possible to adjust the output according to the classification scores; as coarser labels with high confidence are sometimes more useful for agricultural practice than fine-grained but very uncertain labels. We validate the proposed method on a new, large dataset that we make public. ZueriCrop covers an area of 50 km x 48 km in the Swiss cantons of Zurich and Thurgau with a total of 116'000 individual fields spanning 48 crop classes, and 28,000 (multi-temporal) image patches from Sentinel-2. We compare our proposed hierarchical convRNN model with several baselines, including methods designed for imbalanced class distributions. The hierarchical approach performs superior by at least 9.9 percentage points in F1-score.

Optical satellite sensors cannot see the Earth's surface through clouds. Despite the periodic revisit cycle, image sequences acquired by Earth observation satellites are therefore irregularly sampled in time. State-of-the-art methods for crop classification (and other time series analysis tasks) rely on techniques that implicitly assume regular temporal spacing between observations, such as recurrent neural networks (RNNs). We propose to use neural ordinary differential equations (NODEs) in combination with RNNs to classify crop types in irregularly spaced image sequences. The resulting ODE-RNN models consist of two steps: an update step, where a recurrent unit assimilates new input data into the model's hidden state; and a prediction step, in which NODE propagates the hidden state until the next observation arrives. The prediction step is based on a continuous representation of the latent dynamics, which has several advantages. At the conceptual level, it is a more natural way to describe the mechanisms that govern the phenological cycle. From a practical point of view, it makes it possible to sample the system state at arbitrary points in time, such that one can integrate observations whenever they are available, and extrapolate beyond the last observation. Our experiments show that ODE-RNN indeed improves classification accuracy over common baselines such as LSTM, GRU, and temporal convolution. The gains are most prominent in the challenging scenario where only few observations are available (i.e., frequent cloud cover). Moreover, we show that the ability to extrapolate translates to better classification performance early in the season, which is important for forecasting.

We introduce PREDATOR, a model for pairwise point-cloud registration with deep attention to the overlap region. Different from previous work, our model is specifically designed to handle (also) point-cloud pairs with low overlap. Its key novelty is an overlap-attention block for early information exchange between the latent encodings of the two point clouds. In this way the subsequent decoding of the latent representations into per-point features is conditioned on the respective other point cloud, and thus can predict which points are not only salient, but also lie in the overlap region between the two point clouds. The ability to focus on points that are relevant for matching greatly improves performance: PREDATOR raises the rate of successful registrations by more than 20% in the low-overlap scenario, and also sets a new state of the art for the 3DMatch benchmark with 89% registration recall., Comment: CVPR 2021 (Oral) - Improved performance after fixing GNN bug

Continuous observation of climate indicators, such as trends in lake freezing, is important to understand the dynamics of the local and global climate system. Consequently, lake ice has been included among the Essential Climate Variables (ECVs) of the Global Climate Observing System (GCOS), and there is a need to set up operational monitoring capabilities. Multi-temporal satellite images and publicly available webcam streams are among the viable data sources to monitor lake ice. In this work we investigate machine learning-based image analysis as a tool to determine the spatio-temporal extent of ice on Swiss Alpine lakes as well as the ice-on and ice-off dates, from both multispectral optical satellite images (VIIRS and MODIS) and RGB webcam images. We model lake ice monitoring as a pixel-wise semantic segmentation problem, i.e., each pixel on the lake surface is classified to obtain a spatially explicit map of ice cover. We show experimentally that the proposed system produces consistently good results when tested on data from multiple winters and lakes. Our satellite-based method obtains mean Intersection-over-Union (mIoU) scores >93%, for both sensors. It also generalises well across lakes and winters with mIoU scores >78% and >80% respectively. On average, our webcam approach achieves mIoU values of 87% (approx.) and generalisation scores of 71% (approx.) and 69% (approx.) across different cameras and winters respectively. Additionally, we put forward a new benchmark dataset of webcam images (Photi-LakeIce) which includes data from two winters and three cameras., Comment: Accepted for publication in MDPI Remote Sensing Journal

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., Comment: Accepted at IJCV

We investigate active learning in the context of deep neural network models for change detection and map updating. Active learning is a natural choice for a number of remote sensing tasks, including the detection of local surface changes: changes are on the one hand rare and on the other hand their appearance is varied and diffuse, making it hard to collect a representative training set in advance. In the active learning setting, one starts from a minimal set of training examples and progressively chooses informative samples that are annotated by a user and added to the training set. Hence, a core component of an active learning system is a mechanism to estimate model uncertainty, which is then used to pick uncertain, informative samples. We study different mechanisms to capture and quantify this uncertainty when working with deep networks, based on the variance or entropy across explicit or implicit model ensembles. We show that active learning successfully finds highly informative samples and automatically balances the training distribution, and reaches the same performance as a model supervised with a large, pre-annotated training set, with $\approx$99% fewer annotated samples., Comment: 10 pages, 5 figures, ECML/PKDD Workshop on Machine Learning for Earth Observation, 2020

We present a novel 3D shape completion method that operates directly on unstructured point clouds, thus avoiding resource-intensive data structures like voxel grids. To this end, we introduce KAPLAN, a 3D point descriptor that aggregates local shape information via a series of 2D convolutions. The key idea is to project the points in a local neighborhood onto multiple planes with different orientations. In each of those planes, point properties like normals or point-to-plane distances are aggregated into a 2D grid and abstracted into a feature representation with an efficient 2D convolutional encoder. Since all planes are encoded jointly, the resulting representation nevertheless can capture their correlations and retains knowledge about the underlying 3D shape, without expensive 3D convolutions. Experiments on public datasets show that KAPLAN achieves state-of-the-art performance for 3D shape completion., Comment: 18 pages, 15 figures

Most consumer cameras are equipped with electronic rolling shutter, leading to image distortions when the camera moves during image capture. We explore a surprisingly simple camera configuration that makes it possible to undo the rolling shutter distortion: two cameras mounted to have different rolling shutter directions. Such a setup is easy and cheap to build and it possesses the geometric constraints needed to correct rolling shutter distortion using only a sparse set of point correspondences between the two images. We derive equations that describe the underlying geometry for general and special motions and present an efficient method for finding their solutions. Our synthetic and real experiments demonstrate that our approach is able to remove large rolling shutter distortions of all types without relying on any specific scene structure., Comment: CVPR 2020

Modern high-resolution satellite sensors collect optical imagery with ground sampling distances (GSDs) of 30-50cm, which has sparked a renewed interest in photogrammetric 3D surface reconstruction from satellite data. State-of-the-art reconstruction methods typically generate 2.5D elevation data. Here, we present an approach to recover full 3D surface meshes from multi-view satellite imagery. The proposed method takes as input a coarse initial mesh and refines it by iteratively updating all vertex positions to maximize the photo-consistency between images. Photo-consistency is measured in image space, by transferring texture from one image to another via the surface. We derive the equations to propagate changes in texture similarity through the rational function model (RFM), often also referred to as rational polynomial coefficient (RPC) model. Furthermore, we devise a hierarchical scheme to optimize the surface with gradient descent. In experiments with two different datasets, we show that the refinement improves the initial digital elevation models (DEMs) generated with conventional dense image matching. Moreover, we demonstrate that our method is able to reconstruct true 3D geometry, such as facade structures, if off-nadir views are available., Comment: Accepted for publication in ISPRS Journal of Photogrammetry and Remote Sensing

The internal geometry of most modern consumer cameras is not adequately described by the perspective projection. Almost all cameras exhibit some radial lens distortion and are equipped with an electronic rolling shutter that induces distortions when the camera moves during the image capture. When focal length has not been calibrated offline, the parameters that describe the radial and rolling shutter distortions are usually unknown. While for global shutter cameras, minimal solvers for the absolute camera pose and unknown focal length and radial distortion are available, solvers for the rolling shutter were missing. We present the first minimal solutions for the absolute pose of a rolling shutter camera with unknown rolling shutter parameters, focal length, and radial distortion. Our new minimal solvers combine iterative schemes designed for calibrated rolling shutter cameras with fast generalized eigenvalue and Groebner basis solvers. In a series of experiments, with both synthetic and real data, we show that our new solvers provide accurate estimates of the camera pose, rolling shutter parameters, focal length, and radial distortion parameters.

We propose a scheme for supervised image classification that uses privileged information, in the form of keypoint annotations for the training data, to learn strong models from small and/or biased training sets. Our main motivation is the recognition of animal species for ecological applications such as biodiversity modelling, which is challenging because of long-tailed species distributions due to rare species, and strong dataset biases such as repetitive scene background in camera traps. To counteract these challenges, we propose a visual attention mechanism that is supervised via keypoint annotations that highlight important object parts. This privileged information, implemented as a novel privileged pooling operation, is only required during training and helps the model to focus on regions that are discriminative. In experiments with three different animal species datasets, we show that deep networks with privileged pooling can use small training sets more efficiently and generalize better., Comment: Updated version with iNaturalist2018 dataset. privileged pooling, supervised attention, training set bias, fine-grained species recognition, camera trap images

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., Comment: The sequences of the new MOT20 benchmark were previously presented in the CVPR 2019 tracking challenge ( arXiv:1906.04567 ). The differences between the two challenges are: - New and corrected annotations - New sequences, as we had to crop and transform some old sequences to achieve higher quality in the annotations. - New baselines evaluations and different sets of public detections

We present a method to reconstruct the 3D trajectory of an airborne robotic system only from videos recorded with cameras that are unsynchronized, may feature rolling shutter distortion, and whose viewpoints are unknown. Our approach enables robust and accurate outside-in tracking of dynamically flying targets, with cheap and easy-to-deploy equipment. We show that, in spite of the weakly constrained setting, recent developments in computer vision make it possible to reconstruct trajectories in 3D from unsynchronized, uncalibrated networks of consumer cameras, and validate the proposed method in a realistic field experiment. We make our code available along with the data, including cm-accurate ground-truth from differential GNSS navigation., Comment: IROS 2020 camera ready version

Recognising in what type of environment one is located is an important perception task. For instance, for a robot operating in indoors it is helpful to be aware whether it is in a kitchen, a hallway or a bedroom. Existing approaches attempt to classify the scene based on 2D images or 2.5D range images. Here, we study scene recognition from 3D point cloud (or voxel) data, and show that it greatly outperforms methods based on 2D birds-eye views. Moreover, we advocate multi-task learning as a way of improving scene recognition, building on the fact that the scene type is highly correlated with the objects in the scene, and therefore with its semantic segmentation into different object classes. In a series of ablation studies, we show that successful scene recognition is not just the recognition of individual objects unique to some scene type (such as a bathtub), but depends on several different cues, including coarse 3D geometry, colour, and the (implicit) distribution of object categories. Moreover, we demonstrate that surprisingly sparse 3D data is sufficient to classify indoor scenes with good accuracy., Comment: IROS 2020 - Camera Ready

Lake ice is a strong climate indicator and has been recognised as part of the Essential Climate Variables (ECV) by the Global Climate Observing System (GCOS). The dynamics of freezing and thawing, and possible shifts of freezing patterns over time, can help in understanding the local and global climate systems. One way to acquire the spatio-temporal information about lake ice formation, independent of clouds, is to analyse webcam images. This paper intends to move towards a universal model for monitoring lake ice with freely available webcam data. We demonstrate good performance, including the ability to generalise across different winters and different lakes, with a state-of-the-art Convolutional Neural Network (CNN) model for semantic image segmentation, Deeplab v3+. Moreover, we design a variant of that model, termed Deep-U-Lab, which predicts sharper, more correct segmentation boundaries. We have tested the model's ability to generalise with data from multiple camera views and two different winters. On average, it achieves intersection-over-union (IoU) values of ~71% across different cameras and ~69% across different winters, greatly outperforming prior work. Going even further, we show that the model even achieves 60% IoU on arbitrary images scraped from photo-sharing web sites. As part of the work, we introduce a new benchmark dataset of webcam images, Photi-LakeIce, from multiple cameras and two different winters, along with pixel-wise ground truth annotations., Comment: Accepted for ISPRS Congress 2020, Nice, France

Lake ice, as part of the Essential Climate Variable (ECV) lakes, is an important indicator to monitor climate change and global warming. The spatio-temporal extent of lake ice cover, along with the timings of key phenological events such as freeze-up and break-up, provide important cues about the local and global climate. We present a lake ice monitoring system based on the automatic analysis of Sentinel-1 Synthetic Aperture Radar (SAR) data with a deep neural network. In previous studies that used optical satellite imagery for lake ice monitoring, frequent cloud cover was a main limiting factor, which we overcome thanks to the ability of microwave sensors to penetrate clouds and observe the lakes regardless of the weather and illumination conditions. We cast ice detection as a two class (frozen, non-frozen) semantic segmentation problem and solve it using a state-of-the-art deep convolutional network (CNN). We report results on two winters ( 2016 - 17 and 2017 - 18 ) and three alpine lakes in Switzerland. The proposed model reaches mean Intersection-over-Union (mIoU) scores >90% on average, and >84% even for the most difficult lake. Additionally, we perform cross-validation tests and show that our algorithm generalises well across unseen lakes and winters., Comment: Accepted for ISPRS Congress 2020, Nice, France

We propose an embarrassingly simple but very effective scheme for high-quality dense stereo reconstruction: (i) generate an approximate reconstruction with your favourite stereo matcher; (ii) rewarp the input images with that approximate model; (iii) with the initial reconstruction and the warped images as input, train a deep network to enhance the reconstruction by regressing a residual correction; and (iv) if desired, iterate the refinement with the new, improved reconstruction. The strategy to only learn the residual greatly simplifies the learning problem. A standard Unet without bells and whistles is enough to reconstruct even small surface details, like dormers and roof substructures in satellite images. We also investigate residual reconstruction with less information and find that even a single image is enough to greatly improve an approximate reconstruction. Our full model reduces the mean absolute error of state-of-the-art stereo reconstruction systems by >50%, both in our target domain of satellite stereo and on stereo pairs from the ETH3D benchmark., Comment: updated supplementary material

We present a super-resolution method capable of creating a high-resolution texture map for a virtual 3D object from a set of lower-resolution images of that object. Our architecture unifies the concepts of (i) multi-view super-resolution based on the redundancy of overlapping views and (ii) single-view super-resolution based on a learned prior of high-resolution (HR) image structure. The principle of multi-view super-resolution is to invert the image formation process and recover the latent HR texture from multiple lower-resolution projections. We map that inverse problem into a block of suitably designed neural network layers, and combine it with a standard encoder-decoder network for learned single-image super-resolution. Wiring the image formation model into the network avoids having to learn perspective mapping from textures to images, and elegantly handles a varying number of input views. Experiments demonstrate that the combination of multi-view observations and learned prior yields improved texture maps., Comment: 11 pages, 5 figures, 2019 International Conference on 3D Vision (3DV)

The goal of many computer vision systems is to transform image pixels into 3D representations. Recent popular models use neural networks to regress directly from pixels to 3D object parameters. Such an approach works well when supervision is available, but in problems like human pose and shape estimation, it is difficult to obtain natural images with 3D ground truth. To go one step further, we propose a new architecture that facilitates unsupervised, or lightly supervised, learning. The idea is to break the problem into a series of transformations between increasingly abstract representations. Each step involves a cycle designed to be learnable without annotated training data, and the chain of cycles delivers the final solution. Specifically, we use 2D body part segments as an intermediate representation that contains enough information to be lifted to 3D, and at the same time is simple enough to be learned in an unsupervised way. We demonstrate the method by learning 3D human pose and shape from un-paired and un-annotated images. We also explore varying amounts of paired data and show that cycling greatly alleviates the need for paired data. While we present results for modeling humans, our formulation is general and can be applied to other vision problems., Comment: To be published in proceedings of Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI-20)

We present a self-supervised learning framework to estimate the individual object motion and monocular depth from video. We model the object motion as a 6 degree-of-freedom rigid-body transformation. The instance segmentation mask is leveraged to introduce the information of object. Compared with methods which predict dense optical flow map to model the motion, our approach significantly reduces the number of values to be estimated. Our system eliminates the scale ambiguity of motion prediction through imposing a novel geometric constraint loss term. Experiments on KITTI driving dataset demonstrate our system is capable to capture the object motion without external annotation. Our system outperforms previous self-supervised approaches in terms of 3D scene flow prediction, and contribute to the disparity prediction in dynamic area., Comment: Camera Ready Version for CVPRW, 2020

We propose a new STAckable Recurrent cell (STAR) for recurrent neural networks (RNNs), which has fewer parameters than widely used LSTM and GRU while being more robust against vanishing or exploding gradients. Stacking recurrent units into deep architectures suffers from two major limitations: (i) many recurrent cells (e.g., LSTMs) are costly in terms of parameters and computation resources; and (ii) deep RNNs are prone to vanishing or exploding gradients during training. We investigate the training of multi-layer RNNs and examine the magnitude of the gradients as they propagate through the network in the "vertical" direction. We show that, depending on the structure of the basic recurrent unit, the gradients are systematically attenuated or amplified. Based on our analysis we design a new type of gated cell that better preserves gradient magnitude. We validate our design on a large number of sequence modelling tasks and demonstrate that the proposed STAR cell allows to build and train deeper recurrent architectures, ultimately leading to improved performance while being computationally more efficient., Comment: To appear in TPAMI (accepted March 2021)

Up-to-date catalogs of the urban tree population are important for municipalities to monitor and improve quality of life in cities. Despite much research on automation of tree mapping, mainly relying on dedicated airborne LiDAR or hyperspectral campaigns, trees are still mostly mapped manually in practice. We present a fully automated tree detection and species recognition pipeline to process thousands of trees within a few hours using publicly available aerial and street view images of Google MapsTM. These data provide rich information (viewpoints, scales) from global tree shapes to bark textures. Our work-flow is built around a supervised classification that automatically learns the most discriminative features from thousands of trees and corresponding, public tree inventory data. In addition, we introduce a change tracker to keep urban tree inventories up-to-date. Changes of individual trees are recognized at city-scale by comparing street-level images of the same tree location at two different times. Drawing on recent advances in computer vision and machine learning, we apply convolutional neural networks (CNN) for all classification tasks. We propose the following pipeline: download all available panoramas and overhead images of an area of interest, detect trees per image and combine multi-view detections in a probabilistic framework, adding prior knowledge; recognize fine-grained species of detected trees. In a later, separate module, track trees over time and identify the type of change. We believe this is the first work to exploit publicly available image data for fine-grained tree mapping at city-scale, respectively over many thousands of trees. Experiments in the city of Pasadena, California, USA show that we can detect > 70% of the street trees, assign correct species to > 80% for 40 different species, and correctly detect and classify changes in > 90% of the cases.

The success of monocular depth estimation relies on large and diverse training sets. Due to the challenges associated with acquiring dense ground-truth depth across different environments at scale, a number of datasets with distinct characteristics and biases have emerged. We develop tools that enable mixing multiple datasets during training, even if their annotations are incompatible. In particular, we propose a robust training objective that is invariant to changes in depth range and scale, advocate the use of principled multi-objective learning to combine data from different sources, and highlight the importance of pretraining encoders on auxiliary tasks. Armed with these tools, we experiment with five diverse training datasets, including a new, massive data source: 3D films. To demonstrate the generalization power of our approach we use zero-shot cross-dataset transfer}, i.e. we evaluate on datasets that were not seen during training. The experiments confirm that mixing data from complementary sources greatly improves monocular depth estimation. Our approach clearly outperforms competing methods across diverse datasets, setting a new state of the art for monocular depth estimation. Some results are shown in the supplementary video at https://youtu.be/D46FzVyL9I8, Comment: To appear in TPAMI (accepted August 2020)