Image-based shape retrieval (IBSR) aims to retrieve 3D models from a database given a query image, hence addressing a classical task in computer vision, computer graphics, and robotics. Recent approaches typically rely on bridging the domain gap between 2D images and 3D shapes based on the use of multi-view renderings as well as task-specific metric learning to embed shapes and images into a common latent space. In contrast, we address IBSR through large-scale multi-modal pretraining and show that explicit view-based supervision is not required. Inspired by pre-aligned image–point-cloud encoders from ULIP and OpenShape that have been used for tasks such as 3D shape classification, we propose the use of pre-aligned image and shape encoders for zero-shot and standard IBSR by embedding images and point clouds into a shared representation space and performing retrieval via similarity search over compact single-embedding shape descriptors. This formulation allows skipping view synthesis and naturally enables zero-shot and cross-domain retrieval without retraining on the target database. We evaluate pre-aligned encoders in both zero-shot and supervised IBSR settings and additionally introduce a multi-modal hard contrastive loss (HCL) to further increase retrieval performance. Our evaluation demonstrates state-of-the-art performance, outperforming related methods on Acc_Top1 and Acc_Top10 for shape retrieval across multiple datasets, with best results observed for OpenShape combined with Point-BERT. Furthermore, training on our proposed multi-modal HCL yields dataset-dependent gains in standard instance retrieval tasks on shape-centric data, underscoring the value of pretraining and hard contrastive learning for 3D shape retrieval. The code will be made available via the project website.

Visual Place Recognition (VPR) is a core component in computer vision, typically formulated as an image retrieval task for localization, mapping, and navigation. In this work, we instead study VPR as an image pair retrieval front-end for registration pipelines, where the goal is to find top-matching image pairs between two disjoint image sets for downstream tasks such as scene registration, SLAM, and Structure-from-Motion. We comparatively evaluate state-of-the-art VPR families - NetVLAD-style baselines, classification-based global descriptors (CosPlace, EigenPlaces), feature-mixing (MixVPR), and foundation-model-driven methods (AnyLoc, SALAD, MegaLoc) - on three challenging datasets: object-centric outdoor scenes (Tanks and Temples), indoor RGB-D scans (ScanNet-GS), and autonomous-driving sequences (KITTI). We show that modern global descriptor approaches are increasingly suitable as off-the-shelf image pair retrieval modules in challenging scenarios including perceptual aliasing and incomplete sequences, while exhibiting clear, domain-dependent strengths and weaknesses that are critical when choosing VPR components for robust mapping and registration.

Automated and selective harvesting of fruits has become an important area of research, particularly due to challenges such as high costs and a shortage of seasonal labor in advanced economies. This paper focuses on 6D pose estimation of strawberries using purely synthetic data generated through a procedural pipeline for photorealistic rendering. We employ the YOLOX-6D-Pose algorithm, a single-shot approach that leverages the YOLOX backbone, known for its balance between speed and accuracy, and its support for edge inference. To address the lacking availability of training data, we introduce a robust and flexible pipeline for generating synthetic strawberry data from various 3D models via a procedural Blender pipeline, where we focus on enhancing the realism of the synthesized data in comparison to previous work to make it a valuable resource for training pose estimation algorithms. Quantitative evaluations indicate that our models achieve comparable accuracy on both the NVIDIA RTX 3090 and Jetson Orin Nano across several ADD-S metrics, with the RTX 3090 demonstrating superior processing speed. However, the Jetson Orin Nano is particularly suited for resource-constrained environments, making it an excellent choice for deployment in agricultural robotics. Qualitative assessments further confirm the model's performance, demonstrating its capability to accurately infer the poses of ripe and partially ripe strawberries, while facing challenges in detecting unripe specimens. This suggests opportunities for future improvements, especially in enhancing detection capabilities for unripe strawberries (if desired) by exploring variations in color. Furthermore, the methodology presented could be adapted easily for other fruits such as apples, peaches, and plums, thereby expanding its applicability and impact in the field of agricultural automation.

Recent techniques like neural radiance fields (NeRFs) and 3D Gaussian splatting (3DGS) have led to significant improvements in novel view synthesis. Whereas the explicit scene representation of 3DGS in terms of Gaussians allows real-time rendering with state-of-the-art quality, this approach relies on the availability of many views to achieve a coherent scene representation. In this paper, we investigate the importance of accurate camera poses and demonstrate that this even allows for accurate scene representation based on 3D Gaussian Splatting in a sparse-view setting. For this purpose, we address accurate pose estimation by employing a measurement arm equipped with a camera, achieving precise camera-pose estimates with sub-millimeter accuracy. Based on a newly introduced dataset (Core dataset) with its accurate pose information, we demonstrate superior quality in terms of quality of rendered novel views in comparison to results achieved based on calibrations with Dust3R-based and COLMAP-based initializations of the 3D Gaussians. Thereby, our approach offers a reliable and effective solution to practical, sparse-view reconstruction for the preservation of cultural heritage artifacts, which is particularly relevant in applications like virtual museums and archaeology. Furthermore, we expect our Core dataset to serve as a reasonable benchmark, advancing the understanding and development of robust 3D reconstruction methods.

yResulting from the earliest impactful color photography process developed by the Lumiere Brothers, autochromes are invaluable historical artifacts with a high sensitivity and susceptibility to degradation over time. We investigate the potential of deep learning to enhance the experience of historical autochromes through 3D photography and Virtual Reality (VR). Our proposed pipeline utilizes single-image depth estimation and depth-aware inpainting to transform digitized mono-autochromes into layered depth images and corresponding 3D meshes, enabling immersive 3D visualization via a WebXR application. Our user study shows the effectiveness of VR-based experience of autochromes compared to traditional visualization methods, revealing that participants found the VR experience as pleasant as analog exploration despite the latter's authenticity. Thereby, our findings suggest that 3D VR technology could play a crucial role in digitally preserving and revitalizing these culturally significant artifacts and foster investigations with more samples and more types of user interfaces to mature these insights.

We propose a novel cross-spectral rendering framework based on 3D Gaussian Splatting (3DGS) that generates realistic and semantically meaningful splats from registered multi-view spectrum and segmentation maps. This extension enhances the representation of scenes with multiple spectra, providing insights into the underlying materials and segmentation. We introduce an improved physically-based rendering approach for Gaussian splats, estimating reflectance and lights per spectra, thereby enhancing accuracy and realism. In a comprehensive quantitative and qualitative evaluation, we demonstrate the superior performance of our approach with respect to other recent learning-based spectral scene representation approaches (i.e., XNeRF and SpectralNeRF) as well as other non-spectral state-of-the-art learning-based approaches. Our work also demonstrates the potential of spectral scene understanding for precise scene editing techniques like style transfer, inpainting, and removal. Thereby, our contributions address challenges in multi-spectral scene representation, rendering, and editing, offering new possibilities for diverse applications.

The preservation of early visual arts, particularly color photographs, is challenged by deterioration caused by aging and improper storage, leading to issues like blurring, scratches, color bleeding, and fading defects. Despite great advances in image restoration and enhancement in recent years, such systematic defects often cannot be restored by current state-of-the-art software features as available e.g. in Adobe Photoshop, but would require the incorporation of defect-aware priors into the underlying machine learning techniques. However, there are no publicly available datasets of autochromes with defect annotations. In this paper, we address these limitations and present the first approach that allows the automatic removal of greening color defects in digitized autochrome photographs. For this purpose, we introduce an approach for accurately simulating respective defects and use the respectively obtained synthesized data with its ground truth defect annotations to train a generative AI model with a carefully designed loss function that accounts for color imbalances between defected and non-defected areas. As demonstrated in our evaluation, our approach allows for the efficient and effective restoration of the considered defects, thereby overcoming limitations of alternative techniques that struggle with accurately reproducing original colors and may require significant manual effort.

While neural radiance fields (NeRF) led to a breakthrough in photorealistic novel view synthesis, handling mirroring surfaces still denotes a particular challenge as they introduce severe inconsistencies in the scene representation. Previous attempts either focus on reconstructing single reflective objects or rely on strong supervision guidance in terms of additional user-provided annotations of visible image regions of the mirrors, thereby limiting the practical usability. In contrast, in this paper, we present NeRF-MD, a method which shows that NeRFs can be considered as mirror detectors and which is capable of reconstructing neural radiance fields of scenes containing mirroring surfaces without the need for prior annotations. To this end, we first compute an initial estimate of the scene geometry by training a standard NeRF using a depth reprojection loss. Our key insight lies in the fact that parts of the scene corresponding to a mirroring surface will still exhibit a significant photometric inconsistency, whereas the remaining parts are already reconstructed in a plausible manner. This allows us to detect mirror surfaces by fitting geometric primitives to such inconsistent regions in this initial stage of the training. Using this information, we then jointly optimize the radiance field and mirror geometry in a second training stage to refine their quality. We demonstrate the capability of our method to allow the faithful detection of mirrors in the scene as well as the reconstruction of a single consistent scene representation, and demonstrate its potential in comparison to baseline and mirror-aware approaches.

Novel view synthesis (NVS) of urban scenes enables the exploration of cities virtually and interactively, which can further be used for urban planning, navigation, digital tourism, etc. However, many current NVS methods require a large amount of images from known views as input and are sensitive to intrinsic and extrinsic camera parameters. In this paper, we propose a new unified framework for NVS of urban scenes with fewer required views via the integration of scene priors and the joint optimization of camera parameters under an geometric constraint along with NeRF weights. The integration of scene priors makes full use of the priors from the neighbor reference views to reduce the number of required known views. The joint optimization can correct the errors in camera parameters, which are usually derived from algorithms like Structure-from-Motion (SfM), and then further improves the quality of the generated novel views. Experiments show that our method achieves about 25.375 . 375 dB and 25.512 . 512 dB in average in terms of peak signal-to-noise (PSNR) on synthetic and real data, respectively. It outperforms popular state-of-the-art methods (i.e., BungeeNeRF and MegaNeRF ) by about 2-4 4 dB in PSNR. Notably, our method achieves better or competitive results than the baseline method with only one third of the known view images required for the baseline. The code and dataset are available at https://github.com/Dongber/PriNeRF.

We investigate the capabilities of neural inverse procedural modeling to infer high-quality procedural yarn models with fiber-level details from single images of depicted yarn samples. While directly inferring all parameters of the underlying yarn model based on a single neural network may seem an intuitive choice, we show that the complexity of yarn structures in terms of twisting and migration characteristics of the involved fibers can be better encountered in terms of ensembles of networks that focus on individual characteristics. We analyze the effect of different loss functions including a parameter loss to penalize the deviation of inferred parameters to ground truth annotations, a reconstruction loss to enforce similar statistics of the image generated for the estimated parameters in comparison to training images as well as an additional regularization term to explicitly penalize deviations between latent codes of synthetic images and the average latent code of real images in the encoder's latent space. We demonstrate that the combination of a carefully designed parametric, procedural yarn model with respective network ensembles as well as loss functions even allows robust parameter inference when solely trained on synthetic data. Since our approach relies on the availability of a yarn database with parameter annotations and we are not aware of such a respectively available dataset, we additionally provide, to the best of our knowledge, the first dataset of yarn images with annotations regarding the respective yarn parameters. For this purpose, we use a novel yarn generator that improves the realism of the produced results over previous approaches.

Learning-based scene representations such as neural radiance fields or light field networks, that rely on fitting a scene model to image observations, commonly encounter challenges in the presence of inconsistencies within the images caused by occlusions, inaccurately estimated camera parameters or effects like lens flare. To address this challenge, we introduce RANdom RAy Consensus (RANRAC), an efficient approach to eliminate the effect of inconsistent data, thereby taking inspiration from classical RANSAC based outlier detection for model fitting. In contrast to the down-weighting of the effect of outliers based on robust loss formulations, our approach reliably detects and excludes inconsistent perspectives, resulting in clean images without floating artifacts. For this purpose, we formulate a fuzzy adaption of the RANSAC paradigm, enabling its application to large scale models. We interpret the minimal number of samples to determine the model parameters as a tunable hyperparameter, investigate the generation of hypotheses with data-driven models, and analyse the validation of hypotheses in noisy environments. We demonstrate the compatibility and potential of our solution for both photo-realistic robust multi-view reconstruction from real-world images based on neural radiance fields and for single-shot reconstruction based on light-field networks. In particular, the results indicate significant improvements compared to state-of-the-art robust methods for novel-view synthesis on both synthetic and captured scenes with various inconsistencies including occlusions, noisy camera pose estimates, and unfocused perspectives. The results further indicate significant improvements for single-shot reconstruction from occluded images.

3D reconstruction is a long-standing research topic in the photogrammetric and computer vision communities; although a plethora of open-source and commercial solutions for 3D reconstruction have been released in the last few years, several open challenges and limitations still exist. Undoubtedly, deep learning algorithms have demonstrated great potential in several remote sensing tasks, including image-based 3D reconstruction. State-of-the-art monocular and stereo algorithms leverage deep learning techniques and achieve increased performance in depth estimation and 3D reconstruction. However, one of the limitations of such methods is that they highly rely on large training sets that are often tedious to obtain; even when available, they typically refer to indoor, close-range scenarios and low-resolution images. Especially while considering UAV (Unmanned Aerial Vehicle) scenarios, such data are not available and domain adaptation is not a trivial challenge. To fill this gap, the UAV-based multi-sensor dataset for geospatial research (UseGeo - https://usegeo.fbk.eu/home) is introduced in this paper. It contains both image and LiDAR data and aims to support relevant research in photogrammetry and computer vision with a useful training set for both stereo and monocular 3D reconstruction algorithms. In this regard, the dataset provides ground truth data for both point clouds and depth maps. In addition, UseGeo can be also a valuable dataset for other tasks such as feature extraction and matching, aerial triangulation, or image and LiDAR co-registration. The paper introduces the UseGeo dataset and validates some state-of-the-art algorithms to assess their usability for both monocular and multi-view 3D reconstruction.

In cultural heritage, portrait paintings and busts are special genres of artworks which are used to show the appearance and expression of a human subject. The purpose of such artwork is to serve as remembrance of the person who is depicted in that portrait or bust. The bust can moreover serve as a 3D representation of a portrait painting. Therefore, it would be interesting to stylize a portrait painting based on a specific bust, i.e. the generation of a 2D image of a bust corresponding to the person depicted in the portrait image. In this paper, we analyze and discuss the stylization of portrait paintings and photographs of human faces with busts using a deep learning based style transfer approach. To capture the characteristics in the appearance of busts, we created a novel dataset of busts and used DualStyleGAN for the use cases of stylizing portrait paintings and stylizing human faces based on our novel bust style. Our experiments show the potential of this approach. Stylizing human faces as busts might not only be appealing to experts that might save time and effort for generating an initial stylization to refine later on, but also increase the engagement of novice users and exhibition visitors with cultural heritage.

Extracting high-level structural information from 3D point clouds is challenging but essential for tasks like urban planning or autonomous driving requiring an advanced understanding of the scene at hand. Existing approaches are still not able to produce high-quality results consistently while being fast enough to be deployed in scenarios requiring interactivity. We propose to utilize a novel set of features describing the local neighborhood on a per-point basis via first and second order statistics as input for a simple and compact classification network to distinguish between non-edge, sharp-edge, and boundary points in the given data. Leveraging this feature embedding enables our algorithm to outperform the state-of-the-art techniques in terms of quality and processing time.

Despite the impressive progress of telepresence systems for room-scale scenes with static and dynamic scene entities, expanding their capabilities to scenarios with larger dynamic environments beyond a fixed size of a few square-meters remains challenging. In this paper, we aim at sharing 3D live-telepresence experiences in large-scale environments beyond room scale with both static and dynamic scene entities at practical bandwidth requirements only based on light-weight scene capture with a single moving consumer-grade RGB-D camera. To this end, we present a system which is built upon a novel hybrid volumetric scene representation in terms of the combination of a voxel-based scene representation for the static contents, that not only stores the reconstructed surface geometry but also contains information about the object semantics as well as their accumulated dynamic movement over time, and a point-cloud-based representation for dynamic scene parts, where the respective separation from static parts is achieved based on semantic and instance information extracted for the input frames. With an independent yet simultaneous streaming of both static and dynamic content, where we seamlessly integrate potentially moving but currently static scene entities in the static model until they are becoming dynamic again, as well as the fusion of static and dynamic data at the remote client, our system is able to achieve VR-based live-telepresence at close to real-time rates. Our evaluation demonstrates the potential of our novel approach in terms of visual quality, performance, and ablation studies regarding involved design choices.

To enable a richer presentation of cultural heritage and its needs, a shift in how artworks are exhibited is necessary. This paper explores two case studies that highlight the significant role of reproductions in showcasing restoration processes and associated technologies. This approach raises awareness about concepts of care and authenticity and their impact. It goes beyond merely displaying restored digital images that fail to capture the true state of the artworks or the artist’s original intent. To achieve this, we propose employing glass layers or lenticular print technology, allowing a restored version of the artwork’s original state while maintaining the ability to view the original artwork and restoration process separately.

Abstract. Forests are irreplaceable and are being studied extensively. Better forest inventory and understanding necessitate effective mapping, modeling, and automatic analysis. As a result, considerable research effort is being devoted to digitizing forest environments. Recently, digital twins have come to the attention of the geospatial community as a virtual representation of the Earth’s surface linked to its corresponding physical asset. This concept is applicable to forests and has been studied in the literature. This requires initial input data obtained through reality capture. Among mapping techniques, laser scanning has emerged as a state-of-the-art technology for vegetation modeling. In this paper, we look into the potential of mobile laser scanning for forest digital twinning. While most studies concentrate on single tree detection, modeling, and estimation of dendrometric parameters, we also include lower vegetation in our investigations. To accomplish this, we first detect single trees and then investigate different vegetation densities and levels using geometric metrics. We also demonstrate how to model the underlying layers of vegetation in a digital twin. We perform the tests on data from mobile laser scanning (MLS) and compare the results to those from airborne laser scanning (ALS).We show that single tree detection based on crown separation using MLS data works similarly to or slightly better than ALS data. Furthermore, we demonstrate that MLS data allows for more detailed analysis of understory vegetation taking into account different height levels and a multi-level representation, whereas ALS data only allows for rough analysis of the lower parts of forest vegetation.

Partial Differential Equations (PDEs) are notoriously difficult to solve. In general, closed form solutions are not available and numerical approximation schemes are computationally expensive. In this paper, we propose to approach the solution of PDEs based on a novel technique that combines the advantages of two recently emerging machine learning based approaches. First, physics-informed neural networks (PINNs) learn continuous solutions of PDEs and can be trained with little to no ground truth data. However, PINNs do not generalize well to unseen domains. Second, convolutional neural networks provide fast inference and generalize but either require large amounts of training data or a physics-constrained loss based on finite differences that can lead to inaccuracies and discretization artifacts. We leverage the advantages of both of these approaches by using Hermite spline kernels in order to continuously interpolate a grid-based state representation that can be handled by a CNN. This allows for training without any precomputed training data using a physics-informed loss function only and provides fast, continuous solutions that generalize to unseen domains. We demonstrate the potential of our method at the examples of the incompressible Navier-Stokes equation and the damped wave equation. Our models are able to learn several intriguing phenomena such as Karman vortex streets, the Magnus effect, Doppler effect, interference patterns and wave reflections. Our quantitative assessment and an interactive real-time demo show that we are narrowing the gap in accuracy of unsupervised ML based methods to industrial solvers for computational fluid dynamics (CFD) while being orders of magnitude faster.