Layout estimation and 3D object detection are two fundamental tasks in indoor scene understanding. When combined, they enable the creation of a compact yet semantically rich spatial representation of a scene. Existing approaches typically rely on point cloud input, which poses a major limitation since most consumer cameras lack depth sensors and visual-only data remains far more common. We address this issue with TUN3D, the first method that tackles joint layout estimation and 3D object detection in real scans, given multi-view images as input, and does not require ground-truth camera poses or depth supervision. Our approach builds on a lightweight sparse-convolutional backbone and employs two dedicated heads: one for 3D object detection and one for layout estimation, leveraging a novel and effective parametric wall representation. Extensive experiments show that TUN3D achieves state-of-the-art performance across three challenging scene understanding benchmarks: (i) using ground-truth point clouds, (ii) using posed images, and (iii) using unposed images. While performing on par with specialized 3D object detection methods, TUN3D significantly advances layout estimation, setting a new benchmark in holistic indoor scene understanding.

Computer-Aided Design (CAD) plays a central role in engineering and manufacturing, making it possible to create precise and editable 3D models. Using a variety of sensor or user-provided data as inputs for CAD reconstruction can democratize access to design applications. However, most existing methods focus on a single input modality: point clouds, images, or texts, which limits their generalizability and robustness, while few multimodal approaches struggle to deliver competitive quality. Leveraging advances in vision-language models (VLM), we propose $\texttt{cadrille}$, a multimodal CAD reconstruction model that takes inputs of three modalities and outputs executable Python code for CAD reconstruction. Inspired by large language model (LLM) training paradigm, we adopt a two-stage pipeline: supervised fine-tuning (SFT) on large-scale procedurally generated data, followed by reinforcement learning (RL) fine-tuning using online feedback, obtained programatically. In the DeepCAD benchmark, our SFT model outperforms existing single-modal approaches in all three input modalities simultaneously. More importantly, after RL fine-tuning, $\texttt{cadrille}$ sets new state-of-the-art in as many as 10 benchmarks across three modalities and four datasets, including a real-world one.

3D object detection is fundamental for spatial understanding. Real-world environments demand models capable of recognizing diverse, previously unseen objects, which remains a major limitation of closed-set methods. Existing open-vocabulary 3D detectors relax annotation requirements but still depend on training scenes, either as point clouds or images. We take this a step further by introducing Zoo3D, the first training-free 3D object detection framework. Our method constructs 3D bounding boxes via graph clustering of 2D instance masks, then assigns semantic labels using a novel open-vocabulary module with best-view selection and view-consensus mask generation. Zoo3D operates in two modes: the zero-shot Zoo3D_0, which requires no training at all, and the self-supervised Zoo3D_1, which refines 3D box prediction by training a class-agnostic detector on Zoo3D_0-generated pseudo labels. Furthermore, we extend Zoo3D beyond point clouds to work directly with posed and even unposed images. Across ScanNet200 and ARKitScenes benchmarks, both Zoo3D_0 and Zoo3D_1 achieve state-of-the-art results in open-vocabulary 3D object detection. Remarkably, our zero-shot Zoo3D_0 outperforms all existing self-supervised methods, hence demonstrating the power and adaptability of training-free, off-the-shelf approaches for real-world 3D understanding. Code is available at https://github.com/col14m/zoo3d .

3D visual grounding (3DVG) aims to localize objects in a 3D scene based on natural language queries. In this work, we explore zero-shot 3DVG from multi-view images alone, without requiring any geometric supervision or object priors. We introduce Z3D, a universal grounding pipeline that flexibly operates on multi-view images while optionally incorporating camera poses and depth maps. We identify key bottlenecks in prior zero-shot methods causing significant performance degradation and address them with (i) a state-of-the-art zero-shot 3D instance segmentation method to generate high-quality 3D bounding box proposals and (ii) advanced reasoning via prompt-based segmentation, which utilizes full capabilities of modern VLMs. Extensive experiments on the ScanRefer and Nr3D benchmarks demonstrate that our approach achieves state-of-the-art performance among zero-shot methods.

Recent advances in Vision Language Models (VLMs) and Vision Foundation Models (VFMs) have opened new opportunities for zero-shot text-guided segmentation of remote sensing imagery. However, most existing approaches still rely on additional trainable components, limiting their generalisation and practical applicability. In this work, we investigate to what extent text-based remote sensing segmentation can be achieved without additional training, by relying solely on existing foundation models. We propose a simple yet effective approach that integrates contrastive and generative VLMs with the Segment Anything Model (SAM), enabling a fully training-free or lightweight LoRA-tuned pipeline. Our contrastive approach employs CLIP as mask selector for SAM's grid-based proposals, achieving state-of-the-art open-vocabulary semantic segmentation (OVSS) in a completely zero-shot setting. In parallel, our generative approach enables reasoning and referring segmentation by generating click prompts for SAM using GPT-5 in a zero-shot setting and a LoRA-tuned Qwen-VL model, with the latter yielding the best results. Extensive experiments across 19 remote sensing benchmarks, including open-vocabulary, referring, and reasoning-based tasks, demonstrate the strong capabilities of our approach. Code will be released at https://github.com/josesosajs/trainfree-rs-segmentation.

Computer-Aided Design (CAD) plays a foundational role in modern manufacturing and product development, often requiring designers to modify or build upon existing models. Converting 3D scans into parametric CAD representations—a process known as CAD reverse engineering—remains a significant challenge due to the high precision and structural complexity of CAD models. Existing deep learning-based approaches typically fall into two categories: bottom-up, geometry-driven methods, which often fail to produce fully parametric outputs, and top-down strategies, which tend to overlook fine-grained geometric details. Moreover, current methods neglect an essential aspect of CAD modeling: sketch-level constraints. In this work, we introduce a novel approach to CAD reverse engineering inspired by how human designers manually perform the task. Our method leverages multi-plane cross-sections to extract 2D patterns and capture fine parametric details more effectively. It enables the reconstruction of detailed and editable CAD models, outperforming state-of-the-art methods and, for the first time, incorporating sketch constraints directly into the reconstruction process.

Computer-Aided Design (CAD) models are typically constructed by sequentially drawing parametric sketches and applying CAD operations to obtain a 3D model. The problem of 3D CAD reverse engineering consists of reconstructing the sketch and CAD operation sequences from 3D representations such as point clouds. In this paper, we address this challenge through novel contributions across three levels: CAD sequence representation, network design, and training dataset. In particular, we represent CAD sketch-extrude sequences as Python code. The proposed CAD-Recode translates a point cloud into Python code that, when executed, reconstructs the CAD model. Taking advantage of the exposure of pre-trained Large Language Models (LLMs) to Python code, we leverage a relatively small LLM as a decoder for CAD-Recode and combine it with a lightweight point cloud projector. CAD-Recode is trained on a procedurally generated dataset of one million CAD sequences. CAD-Recode significantly outperforms existing methods across the DeepCAD, Fusion360 and real-world CC3D datasets. Furthermore, we show that our CAD Python code output is interpretable by off-the-shelf LLMs, enabling CAD editing and CAD-specific question answering from point clouds.

Multi-modal data in Earth Observation (EO) presents a huge opportunity for improving transfer learning capabilities when pre-training deep learning models. Unlike prior work that often overlooks multi-modal EO data, recent methods have started to include it, resulting in more effective pre-training strategies. However, existing approaches commonly face challenges in effectively transferring learning to downstream tasks where the structure of available data differs from that used during pre-training. This paper addresses this limitation by exploring a more flexible multi-modal, multi-task pre-training strategy for EO data. Specifically, we adopt a Multi-modal Multi-task Masked Autoencoder (MultiMAE) that we pre-train by reconstructing diverse input modalities, including spectral, elevation, and segmentation data. The pre-trained model demonstrates robust transfer learning capabilities, outperforming state-of-the-art methods on various EO datasets for classification and segmentation tasks. Our approach exhibits significant flexibility, handling diverse input configurations without requiring modality-specific pre-trained models. Code will be available at: https://github.com/josesosajs/multimae-meets-eo.

Growing customer demand for smart solutions in robotics and augmented reality has attracted considerable attention to 3D object detection from point clouds. Yet, existing indoor datasets taken individually are too small and insufficiently diverse to train a powerful and general 3D object detection model. In the meantime, more general approaches utilizing foundation models are still inferior in quality to those based on supervised training for a specific task. In this work, we propose UniDet3D, a simple yet effective 3D object detection model, which is trained on a mixture of indoor datasets and is capable to work in various indoor environments. By unifying different label spaces, UniDet3D enables learning a strong representation across multiple datasets through a supervised joint training scheme. The proposed network architecture is built upon a vanilla transformer encoder, making it easy to run, customize and extend the prediction pipeline for practical use. Extensive experiments demonstrate that UniDet3D obtains significant gains over existing 3D object detection methods in 6 indoor benchmarks: ScanNet (+1.1 mAP50), S3DIS (+9.1 mAP50), ARKitScenes (+19.4 mAP25), MultiScan (+14.3 mAP50), 3RScan (+3.2 mAP50), and ScanNet++ (+2.7 mAP50).

In this paper the authors analyzed data from five different sources – topographic maps, land management materials, UAV survey, retrospective monitoring of soil and land cover, and recognition of open soil surface - to identify and involve abandoned arable land into agricultural circulation. The result of “on-ground” verification showed the different accuracy of the methods. Retrospective monitoring of soil and land cover provides the highest correctness, close to 100%. The combination of this technology with recognition of open soil surface, which allows cutting off unsuitable lands, can provide a solution for inventory arable land and introducing it into agricultural circulation.

The results of a field study of the erosion–accumulative patterns of the soil cover of a key area in the zone of dark chestnut soils (Kastanozems) are presented, and the information content of multitemporal remote sensing data on the bare soil surface for its identification and mapping is analyzed. The site is located on the Millerovo–Morozovskaya inclined plain within the Don–Donetsk hilly-ridge plain, in Oblivsky district of the Rostov region. The soil cover of the key area is represented by a combination of low-contrasting soils on convex and concave surfaces within an elongated ridge and its slopes, including agro-dark chestnut solonetzic and nonsolonetzic soils, agrozems (washed away soils that have lost the middle-profile xerometamorphic horizon), and stratozems (aggraded soils). The development of this soil cover pattern is determined by a combination of mesorelief landforms and two types of soil-forming rocks. The map of the C coefficient of the multitemporal soil line reveals the heterogeneity of the soil cover related to the activity of erosion/deposition processes. In the key area, three groups of contrasting soil combinations differ significantly in the form of different variations and combinations-variations forming a kind of framework of the soil cover. Combinations of eroded and aggraded soils located between the above three groups of soil combinations significantly differ from their neighbors, but their interpretation has an increased uncertainty.

Current medium- and small-scale estimates of soil erosion in Russia are very few. At the same time, a favorable situation has now developed for assessing the rates and volumes of soil erosion losses. Erosion models adapted to available digital elevation models, various farmland maps, and climate databases are now available to researchers. We have estimated the rates and volumes of soil erosion with the use of different maps of cropland: world cover from the European Space Agency (ESA WC), global land cover and land use from the University of Maryland (GLCLU), the official unified federal information system of agricultural land (UFIS AL) from the Ministry of Agriculture of the Russian Federation, and the original conventionally reference land cover map of Alekseevskii district of Belgorod oblast. It has been found that the UFIS AL map gives cropland areas close to the average from first three maps. Public access maps (ESA WC and GLCLU) give maximum and minimum estimates of cropland, respectively. A comparison with the conventionally reference large-scale map shows that the accuracy of the UFIS AL does not exceed 90

Adoption of the farming systems that aim to sequester carbon in agricultural soils is one of the ways to mitigate global climate change. This study focuses on the estimation of organic carbon sequestration potential of the Russian croplands in the upper (0–30 cm) soil layer by creating a set of maps using the data from global and national databases as the input data. The maps are generated within the FAO Global Soil Organic Carbon Sequestration Potential Map (GSOCseq) project according to the unified methodology using the RothC model to predict the rate of carbon sequestration in 2020–2040 under a business as usual scenario (BAU), as well as under sustainable soil management scenarios with additional different C input (+5, +10, and +20

Current medium- and small-scale estimates of soil erosion in Russia are very few. At the same time, a favorable situation has now developed for assessing the rates and volumes of soil erosion losses. Erosion models have been developed that are adapted to available digital elevation models, various farmland masks and climate databases have been created. The paper studies the accuracy of erosion estimates using various maps of arable land. Two maps are public (ESA WC, GLCLU), the third is the official Ministry of Agriculture (MA) of the Russian Federation, the fourth map is an author’s reference map of Alekseevsky district. It has been established that the map of the MA gives the most average arable land areas among the first three maps. Public access maps showed maximum and minimum estimates of arable land area. Comparison with the standard showed that the accuracy of the map of the MA does not exceed 90%, the remaining maps – 84 and 83%. The area of arable land in the Belgorod region varies slightly (from 1,445 to 1,586 thousand hectares); so the region is favorable for erosion modelling. Deviations from the average rates of soil erosion calculated using different maps of arable land in the region as a whole amounted to 7%, and in some areas reached 27%. Thus, today assessments of soil erosion at the regional level can be carried out with an error of at least 10–15% only as a result of the uncertainty in mapping the boundaries of arable land. In the Russia as a whole, data on the area of arable land varies significantly, from 80 to 132 million hectares. Consequently, the use of existing maps of arable land can lead to significant uncertainties in soil erosion estimates averaged at the level of districts and above.

Most 3D instance segmentation methods exploit a bottom-up strategy, typically including resource-exhaustive post-processing. For point grouping, bottom-up methods rely on prior assumptions about the objects in the form of hyperparameters, which are domain-specific and need to be carefully tuned. On the contrary, we address 3D instance segmentation with a TD3D: the pioneering cluster-free, fully-convolutional and entirely data-driven approach trained in an end-to-end manner. This is the first top-down method outperforming bottom-up approaches in 3D domain. With its straightforward pipeline, it demonstrates outstanding accuracy and generalization ability on the standard indoor benchmarks: ScanNet v2, its extension ScanNet200, and S3DIS, as well as on the aerial STPLS3D dataset. Besides, our method is much faster on inference than the current state-of-the-art grouping-based approaches: our flagship modification is 1.9x faster than the most accurate bottom-up method, while being more accurate, and our faster modification shows state-of-the-art accuracy running at 2.6x speed. Code is available at https://github.com/SamsungLabs/td3d .

Semantic, instance, and panoptic segmentation of 3D point clouds have been addressed using task-specific models of distinct design. Thereby, the similarity of all segmentation tasks and the implicit relationship between them have not been utilized effectively. This paper presents a unified, simple, and effective model addressing all these tasks jointly. The model, named OneFormer3D, performs instance and semantic segmentation consistently, using a group of learnable kernels, where each kernel is responsible for generating a mask for either an instance or a semantic category. These kernels are trained with a transformer-based decoder with unified instance and semantic queries passed as an input. Such a design enables training a model end-to-end in a single run, so that it achieves top performance on all three segmentation tasks simultaneously. Specifically, our OneFormer3D ranks 1st and sets a new state-of-the-art (+2.1 mAP50) in the ScanNet test leaderboard. We also demonstrate the state-of-the-art results in semantic, instance, and panoptic segmentation of ScanNet (+21 PQ), ScanNet200 (+3.8 mAP50), and S3DIS (+0.8 mIoU) datasets.

For the test monitoring site on the territory of the Oroshaemaya Experimental Station in the south of the Volga Upland (Volgograd oblast, Russia), the estimates of the baseline contents and stocks of soil organic carbon (Corg) and soil bulk density and characteristics of their spatial variability and vertical distribution patterns within the upper 40 cm encompassing the entire humus layer are presented. The specified indicators characterize arable land in crop rotation under dry and irrigated farming practices and a 50-year-old fallow at seven dynamic monitoring plots with light chestnut soils. The studied soils have two types of profile horizonation: P–BMK–BCAnc (agrochestnut soils) and Pca–BCAnc (carbonate-accumulative agrozems); they are silty loamy and develop from yellowish brown sandy silt loams. In the WRB system, they are classified as Haplic Kastanozems (Loamic, Aric). The map of the coefficients of multitemporal soil line, retrospective analysis of remote sensing data for 40 years, and the results of field soil survey have been used to select the monitoring plots. Each plot 30 × 30 m includes nine individual sampling points, at which soil sampling from thin (5 or 10 cm) layers has been performed to a depth of 40 cm. The coefficient of spatial variation in the Corg content increases from 5–10 to 30–40

Recently, sparse 3D convolutions have changed 3D object detection. Performing on par with the voting-based approaches, 3D CNNs are memory-efficient and scale to large scenes better. However, there is still room for improvement. With a conscious, practice-oriented approach to problem-solving, we analyze the performance of such methods and localize the weaknesses. Applying modifications that resolve the found issues one by one, we end up with TR3D: a fast fully-convolutional 3D object detection model trained end-to-end, that achieves state-of-the-art results on the standard benchmarks, ScanNet v2, SUN RGB-D, and S3DIS. Moreover, to take advantage of both point cloud and RGB inputs, we introduce an early fusion of 2D and 3D features. We employ our fusion module to make conventional 3D object detection methods multimodal and demonstrate an impressive boost in performance. Our model with early feature fusion, which we refer to as TR3D+FF, outperforms existing 3D object detection approaches on the SUN RGB-D dataset. Overall, besides being accurate, both TR3D and TR3D+FF models are lightweight, memory-efficient, and fast, thereby marking another milestone on the way toward real-time 3D object detection. Code is available at https://github.com/SamsungLabs/tr3d.

— According to published and archival data, the dynamics of soil plowing in Tula oblast in general and in its particular districts has been traced. It has been found that the active agricultural development in the southern part of the modern Tula oblast began no earlier than in the middle of the 17th century. According to the maps of different years, changes in the area of arable land in the Lokna River basin (Plavsk district of Tula oblast) from the end of the 18th century until the present time have been determined. The reliability of our data on the history of plowing of the territory is confirmed by the literature data. The composition of field crops and changes in farming systems during the period under review have been reconstructed, and a significant conservatism of these indicators has been revealed: the fallow three-field system dominated throughout the 19th century and the first half of the 20th century; the predominance of rye among winter crops and oats among spring crops remained until the 1970s. According to the WaTEM/SEDEM and using historical data on changes in the main factors of soil erosion as input parameters, the intensity and volumes of rainfall runoff for the selected stages of development of the Lokna River basin have been calculated. In order to compare the results with the literature data for the period of 1980–1993, the rates and volumes of snowmelt washout have been estimated. An assessment of the reliability of the reconstructed model parameters for remote historical epochs has been carried out. The analysis of historical changes in the factors and rates of erosion indicates that the maximum extent of plowed area and the maximum intensity of erosional processes in Lokna River took place in the last quarter of the 19th–early 20th centuries. The influence of particular geomorphic positions of arable land on the rates and volumes of soil erosion has been revealed.

. Recently, promising applications in robotics and augmented reality have attracted considerable attention to 3D object detection from point clouds. In this paper, we present FCAF3D — a first-in-class fully convolutional anchor-free indoor 3D object detection method. It is a simple yet effective method that uses a voxel representation of a point cloud and processes voxels with sparse convolutions. FCAF3D can handle large-scale scenes with minimal runtime through a single fully convolutional feed-forward pass. Existing 3D object detection methods make prior assumptions on the geometry of objects, and we argue that it limits their generalization ability. To eliminate prior assumptions, we propose a novel parametrization of oriented bounding boxes that allows obtaining better results in a purely data-driven way. The proposed method achieves state-of-the-art 3D object detection results in terms of mAP@0.5 on ScanNet V2 ( +4.5 ), SUN RGB-D ( +3.5 ), and S3DIS ( +20.5 ) datasets. The code and models are available at https://github.com/samsunglabs/fcaf3d .