Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Hong SunSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works391 from public data
- Deep Learning in Remote Sensing: A Comprehensive Review and List of Resources3,334
The challenges of using deep learning for remote sensing data analysis are analyzed, the recent advances are reviewed, and resources are provided to make deep learning in remote sensing ridiculously simple to start with.
- DOTA: A Large-Scale Dataset for Object Detection in Aerial Images3,266
A large-scale Dataset for Object deTection in Aerial images (DOTA) is introduced and state-of-the-art object detection algorithms on DOTA are evaluated, demonstrating that DOTA well represents real Earth Vision applications and are quite challenging.
- AID: A Benchmark Data Set for Performance Evaluation of Aerial Scene Classification2,351
The Aerial Image data set (AID), a large-scale data set for aerial scene classification, is described to advance the state of the arts in scene classification of remote sensing images and can be served as the baseline results on this benchmark.
- 1,607
- Align Deep Features for Oriented Object Detection1,221
A single-shot alignment network (S2A-Net) consisting of two modules: a feature alignment module (FAM) and an oriented detection module (ODM) that can achieve the state-of-the-art performance on two commonly used aerial objects’ data sets while keeping high efficiency.
- Transferring Deep Convolutional Neural Networks for the Scene Classification of High-Resolution Remote Sensing Imagery1,195
This paper proposes two scenarios for generating image features via extracting CNN features from different layers and reveals that the features from pre-trained CNNs generalize well to HRRS datasets and are more expressive than the low- and mid-level features.
- Land-cover classification with high-resolution remote sensing images using transferable deep models1,022
The main idea is to rely on deep neural networks for presenting the contextual information contained in different types of land-covers and propose a pseudo-labeling and sample selection scheme for improving the transferability of deep models.
- Gliding Vertex on the Horizontal Bounding Box for Multi-Oriented Object Detection1,013
An obliquity factor based on area ratio between the object and its horizontal bounding box, guiding the selection of horizontal or oriented detection for each object is introduced, and five extra target variables are added to the regression head of faster R-CNN, which requires ignorable extra computation time.
- Remote Sensing Image Scene Classification Meets Deep Learning: Challenges, Methods, Benchmarks, and Opportunities994
This article provides a systematic survey of deep learning methods for remote sensing image scene classification by covering more than 160 papers and discusses the main challenges of remote sensing images classification and survey.
- ReDet: A Rotation-equivariant Detector for Aerial Object Detection935
A Rotation-equivariant Detector (ReDet) is proposed, which explicitly encodes rotation equivariance and rotation invariance and incorporates rotation- equivariant networks into the detector to extract rotation-Equivariant features, which can accurately predict the orientation and lead to a huge reduction of model size.
- Object Detection in Aerial Images: A Large-Scale Benchmark and Challenges714
The expanded large-scale DOTA dataset, the extensive baselines, the code library and the challenges can facilitate the designs of robust algorithms and reproducible research on the problem of object detection in aerial images.
- Rotation-Sensitive Regression for Oriented Scene Text Detection563
The proposed method named Rotation-sensitive Regression Detector (RRD) achieves state-of-the-art performance on several oriented scene text benchmark datasets, including ICDAR 2015, MSRA-TD500, RCTW-17, and COCO-Text, and achieves a significant improvement on a ship collection dataset, demonstrating its generality on oriented object detection.
- UAVid: A semantic segmentation dataset for UAV imagery514
The UAVid dataset is introduced, a new high-resolution UAV semantic segmentation dataset as a complement, which brings new challenges, including large scale variation, moving object recognition and temporal consistency preservation.
- Detecting tiny objects in aerial images: A normalized Wasserstein distance and a new benchmark461
A new evaluation metric dubbed Normalized Wasserstein Distance (NWD) and a new RanKing-based Assigning (RKA) strategy for tiny object detection are proposed, which can be easily embedded into all kinds of anchor-based detectors to replace the standard IoU threshold-based one.
- A survey on vision-based UAV navigation447
This article is to present a comprehensive literature review of the vision-based methods for UAV navigation, specifically on visual localization and mapping, obstacle avoidance and path planning, which compose the essential parts of visual navigation.
- 417
- 406
- 381
- 361
- STRUCTURAL HIGH-RESOLUTION SATELLITE IMAGE INDEXING352
A satellite image indexing method relying on topographic maps and a shape-based image indexed scheme that is robust to changes in scale, orientation and contrast is suggested.
- An Empirical Study of Remote Sensing Pretraining321
Empirical study shows that RSP can help deliver distinctive performances in scene recognition tasks and in perceiving RS related semantics such as “Bridge” and “Airplane”, and finds that, although RSP mitigates the data discrepancies of traditional ImageNet pretraining on RS images, it may still suffer from task discrepancies, where downstream tasks require different representations from scene Recognition tasks.
- Decoupling Zero-Shot Semantic Segmentation319
A simple and effective zero-shot semantic segmentation model, called ZegFormer, is proposed, which outperforms the previous methods on ZS3 standard benchmarks by large margins, e.g., 22 points on the PAS-CAL VOC and 3 points onThe COCO-Stuff in terms of mIoU for unseen classes.
- 318
- 308
- On Creating Benchmark Dataset for Aerial Image Interpretation: Reviews, Guidances, and Million-AID292
The current challenges of developing intelligent algorithms for RS image interpretation with bibliometric investigations are analyzed and the general guidances on creating benchmark datasets in efficient manners are presented.
- Multi-feature combined cloud and cloud shadow detection in GaoFen-1 wide field of view imagery288
An automatic multi-feature combined (MFC) method is proposed for cloud and cloud shadow detection in GF-1 WFV imagery over land and results indicate that MFC performs well under different land conditions, and the average cloud classification accuracy of MFC is as high as 98.3%.
- 273
- 257
- 189
- Dynamic Coarse-to-Fine Learning for Oriented Tiny Object Detection181
A dynamic prior along with the coarse-to-fine assigner is proposed, dubbed DCFL, which leverages the coarse prior matching and finer posterior constraint to dynamically assign labels, providing appropriate and relatively balanced supervision for diverse instances.
- GLF-CR: SAR-enhanced cloud removal with global–local fusion179
A novel global-local fusion based cloud removal (GLF-CR) algorithm to leverage the complementary information embedded in SAR images to generate reliable texture details of the missing regions and outperform state-of-the-art cloud removal algorithms.
- An Urban Water Extraction Method Combining Deep Learning and Google Earth Engine174
This article proposed a new method by combining Google Earth Engine (GEE) with a multiscale convolutional neural network (MSCNN) to extract urban water from Landsat images, which can be summarized as “offline training and online prediction” (OTOP).
- 168
- Multi-feature combined cloud and cloud shadow detection in GF-1 WFV imagery.166
An automatic multi-feature combined (MFC) method is proposed for cloud and cloud shadow detection in GF-1 WFV imagery over land and results indicate that MFC performs well under different land conditions, and the average cloud classification accuracy of MFC is as high as 98.3%.
- Multi-feature combined cloud and cloud shadow detection in GF-1 WFV imagery over land166
An automatic multi-feature combined (MFC) method is proposed for cloud and cloud shadow detection in GF-1 WFV imagery over land and results indicate that MFC performs well under different land conditions, and the average cloud classification accuracy of MFC is as high as 98.3%.
- 164
- Learning Attraction Field Representation for Robust Line Segment Detection161
A region-partition based attraction field dual representation for line segment maps, which poses the problem of line segment detection (LSD) as the region coloring problem and harnesses the best practices developed in ConvNets based semantic segmentation methods such as the encoder-decoder architecture and the a-trous convolution.
- Object Tracking in Satellite Videos by Improved Correlation Filters With Motion Estimations161
This article addresses the problem of fast object tracking in satellite videos, by developing a novel tracking algorithm based on correlation filters embedded with motion estimations, based on the kernelized correlation filter (KCF).
- Event Enhanced High-Quality Image Recovery156
An explainable network, an event-enhanced sparse learning network (eSL-Net), to recover the high-quality images from event cameras and can largely improve the performance of the state-of-the-art by 7-12 dB.
- A Multiple-Instance Densely-Connected ConvNet for Aerial Scene Classification151
A multiple-instance densely-connected ConvNet (MIDC-Net) for aerial scene classification is proposed that outperforms many state-of-the-art methods by a large margin with much fewer parameters.
- Mini-UAV-based Remote Sensing: Techniques, Applications and Prospectives145
UAVs have many promising characteristics—flexibility, efficiency, high spatial/temporal resolution, low cost, easy operation, and so forth—that make them an effective complement to the other two platforms and a cost-effective means for remote sensing.
- 140
- Holistically-Attracted Wireframe Parsing138
This paper presents a fast and parsimonious parsing method to accurately and robustly detect a vectorized wireframe in an input image with a single forward pass, and is thus called Holistically-Attracted Wireframe Parser (HAWP).
- 137
- Parsing very high resolution urban scene images by learning deep ConvNets with edge-aware loss135
A standalone end-to-end edge-aware neural network (EaNet) is proposed for urban scene semantic segmentation that incorporates a large kernel pyramid pooling (LKPP) module to capture rich multi-scale context with strong continuous feature relations.
- Exploiting Deep Features for Remote Sensing Image Retrieval: A Systematic Investigation132
This article analyzes the three core issues of RS image retrieval and provides a comprehensive review on existing methods and focuses on the feature extraction issue and delve how to use powerful deep representations to address this task.
- 129
- Accurate Junction Detection and Characterization in Natural Images123
A generic junction analysis scheme that yields a characterization of L-, Y- and X- junctions, including a precise computation of their type, localization and scale, using a statistical modeling of suitably normalized gray level gradients.
- Motion Deblurring with Real Events120
An end-to-end learning framework for event-based motion deblurring in a self-supervised manner, where real-world events are exploited to alleviate the performance degradation caused by data inconsistency, and a piecewise linear motion model is proposed to take into account motion non-linearities.
- Image stitching by line-guided local warping with global similarity constraint118
A line-guided local warping method with a global similarity constraint for image stitching that provides a convincing stitching performance and that it outperforms other state-of-the-art methods.
- 117
- Detecting Power Lines in UAV Images with Convolutional Features and Structured Constraints114
The proposed method fully exploits multiscale and structured prior information to conduct both accurate and efficient detection and achieves competitive performance compared with state-of-the-art methods.
- Unmixing Convolutional Features for Crisp Edge Detection110
A novel tracing loss that performs feature unmixing by tracing boundaries for better side edge learning, and a context-aware fusion block that tackles the side mixing by aggregating the complementary merits of learned side edges.
- 105
- 102
- Generating annual high resolution land cover products for 28 metropolises in China based on a deep super-resolution mapping network using Landsat imagery100
The proposed method can facilitate a seamless and fine-grained observation system that can fill the weakness of the existing land cover activities and provide a brand-new insight into the urban dynamic and its underlying mechanism.
- 97
- Expanding Low-Density Latent Regions for Open-Set Object Detection95
This work proposes to identify unknown objects by separating high/low-density regions in the latent space, based on the consensus that unknown objects are usually distributed in low-density latent regions, and presents a novel Openset Detector (OpenDet) with expanded low- density regions.
- A geometry-attentional network for ALS point cloud classification94
A geometry-attentional network consisting of geometry-aware convolution, dense hierarchical architecture and elevation-att attention module to embed the three characteristics of ALS point cloud classification effectively is proposed, which can be trained in an end-to-end manner.
- 90
- Learning to Calibrate Straight Lines for Fisheye Image Rectification90
This paper proposes a novel deep neural network to impose explicit geometry constraints onto processes of the fisheye lens calibration and the distorted image rectification, and achieves the best published rectification quality and the most accurate estimation of distortion parameters on a large set of synthetic and real f isheye images.
- Parsing Table Structures in the Wild86
This paper aims to establish a practical table structure parsing system for real-world scenarios where tabular input images are taken or scanned with severe deformation, bending or occlusions and proposes an approach named Cycle-CenterNet on the top of CenterNet with a novel cycle-pairing module to simultaneously detect and group tabular cells into structured tables.
- FGN: Fully Guided Network for Few-Shot Instance Segmentation84
A Fully Guided Network for few-shot instance segmentation is presented, which introduces different guidance mechanisms into the various key components in Mask R-CNN in order to make full use of the guidance effect from the support set and adapt better to the inter-class generalization.
- 83
- 82
- HGFormer: Hierarchical Grouping Transformer for Domain Generalized Semantic Segmentation79
Experiments show that HGFormer yields more robust semantic segmentation results than per-pixel classification methods and flat-grouping transformers, and outperforms previous methods significantly.
- 79
- Image Caption Generation with Part of Speech Guidance78
Experimental results on the most popular benchmark datasets, e.g., Flickr30k and MS COCO, consistently demonstrate that the method can significantly enhance the performance of a standard image caption generation model, and achieve the conpetitive results.
- Mid-level features and spatio-temporal context for activity recognition72
A framework that computes mid-level features and takes into account their contextual information, and introduces a spatio-temporal context kernel (STCK), which not only captures local properties of features but also considers their spatial and temporal context information.
- A Hierarchical Scheme of Multiple Feature Fusion for High-Resolution Satellite Scene Categorization71
Experimental analysis show that the proposed classification model based on a hierarchical fusion of multiple features based on support vector machine with histogram intersection kernel leads to state-of-the-art classification performance on the satellite scenes.
- 70
- A Color-Texture-Structure Descriptor for High-Resolution Satellite Image Classification67
This paper proposes combining the two methods under a tree of regions and presents a new descriptor to encode color, texture and structure features using a hierarchical structure-Color Binary Partition Tree (CBPT), which is called the CTS descriptor.
- Synthesizing and Mixing Stationary Gaussian Texture Models65
Two classes of Gaussian processes parameterized by a small number of compactly supported linear filters, the so-called textons, provide state of the art results on stationary dynamical texture synthesis and mixing.
- Unleashing Unlabeled Data: A Paradigm for Cross-View Geo-Localization64
An unsupervised framework including a cross-view projection to guide the model for retrieving initial pseudo-labels and a fast re-ranking mechanism to refine the pseudo-labels by leveraging the fact that “the perfectly paired ground-satellite image is located in a unique and identical scene” is proposed.
- Learning to Extract Building Footprints From Off-Nadir Aerial Images62
An offset vector learning scheme is proposed, which turns the building footprint extraction problem in off-nadir images into an instance-level joint prediction problem of the building roof and its corresponding roof to footprint offset vector, and a simple but effective feature-level offset augmentation module is proposed.
- NaSC-TG2: Natural Scene Classification With Tiangong-2 Remotely Sensed Imagery61
A large-scale dataset named “NaSC-TG2,” which is a novel benchmark dataset for remote sensing natural scene classification built from Tiangong-2 remotely sensed imagery, is described, which offers natural scenes with novel spatial scale and imaging performance compared with other datasets.
- Texture synthesis through convolutional neural networks and spectrum constraints59
This paper presents a significant improvement for the synthesis of texture images using convolutional neural networks (CNNs), making use of constraints on the Fourier spectrum of the results.
- 58
- Anomaly Detection in Aerial Videos With Transformers57
A new baseline model, anomaly detection with Transformers (ANDTs), is presented, which treats consecutive video frames as a sequence of tubelets, utilizes a Transformer encoder to learn feature representations from the sequence, and leverages a decoder to predict the next frame.
- All Grains, One Scheme (AGOS): Learning Multigrain Instance Representation for Aerial Scene Classification55
This article proposes a novel all grains, one scheme (AGOS) framework, which is the first work to extend the classic multiple instance learning (MIL) into multigrain formulation and achieves a comparable performance against the state-of-the-art methods.
- SegEarth-R1: Geospatial Pixel Reasoning via Large Language Model53
SegEarth-R1 is proposed, a simple yet effective language-guided segmentation baseline that integrates a hierarchical visual encoder, a large language model (LLM) for instruction parsing, and a tailored mask generator for spatial correlation, and significantly outperforming traditional and LLM-based segmentation methods.
- PlaneTR: Structure-Guided Transformers for 3D Plane Recovery52
A neural network built upon Transformers, namely PlaneTR, to simultaneously detect and reconstruct planes from a single image and achieves state-of-the-art performance on the ScanNet and NYUv2 datasets.
- Vision-Language Modeling Meets Remote Sensing: Models, datasets, and perspectives50
This article covers a taxonomy of VLM in remote sensing: contrastive learning, visual instruction tuning, and text-conditioned image generation, and a thorough review of existing works, examining foundation models and task-specific adaptation methods in contrastive-based VLM.
- 50
- Detecting building changes with off-nadir aerial images49
A multi-task guided change detection network model, named as MTGCD-Net, designed to tackle the building roof mismatch problem with off-nadir aerial images achieves superior performance over the previous state-of-the-art competitors.
- 49
- Accurate Annotation of Remote Sensing Images via Active Spectral Clustering with Little Expert Knowledge49
This paper introduces a novel active clustering method for the annotation of high-resolution remote sensing images that can achieve high accuracy in the task of remote sensing image annotation with relatively little expert knowledge, thereby greatly lightening the workload burden and reducing the requirements regarding expert knowledge.
- Learning Transferable Deep Models for Land-Use Classification with High-Resolution Remote Sensing Images.48
The main idea is to rely on deep neural networks for presenting the semantic information contained in different types of land-uses and propose a pseudo-labeling and sample selection scheme for improving the transferability of deep models.
- Extreme value theory-based calibration for the fusion of multiple features in high-resolution satellite scene classification48
A hierarchical classification method for high-resolution satellite imagery incorporating extreme value theory (EVT)-based normalization to calibrate multiple-feature scores and achieves promising classification results and outperforms other competitive methods.
- Meaningful Object Segmentation From SAR Images via a Multiscale Nonlocal Active Contour Model47
A novel SAR image segmentation method by exploiting a multiscale active contour model based on the nonlocal processing principle that can not only achieve precise segmentations for images with heavy speckle and nonlocal intensity variations but also be used for SAR images from different types of sensors.
- Revisiting Document Image Dewarping by Grid Regularization46
This paper first learns the boundary points and the pixels in the text lines and then follows the most simple observation that the boundaries and text lines in both horizontal and vertical directions should be kept after dewarping to introduce a novel grid regularization scheme.
- 46
- 45
- Learning Local-Global Contextual Adaptation for Multi-Person Pose Estimation45
This paper proposes a multi-person pose estimation approach, dubbed as LOGO-CAP, by learning the LOcal-GlObal Contextual Adaptation for human Pose, which is end-to-end trainable with near real-time inference speed in a single forward pass, obtaining state-of-the-art performance on the COCO keypoint benchmark for bottom-up human pose estimation.
- 45
- Deep Graph Matching under Quadratic Constraint44
This paper proposes to explicitly formulate pairwise graph structures as a quadratic constraint incorporated into the DGM framework, which minimizes the pairwise structural discrepancy between graphs, which can reduce the ambiguities brought by only using the extracted CNN features.
- Automatic Ship Detection in SAR Images Using Multi-Scale Heterogeneities and an A Contrario Decision44
A new ship detection approach based on multi-scale heterogeneities under the a contrario decision framework, with a few parameters that can be easily determined, which works more stably for ships with different sizes, and has better performance than conventional ship detectors.
- Event-based Synthetic Aperture Imaging with a Hybrid Network43
A novel SAI system based on the event camera which can produce asynchronous events with extremely low latency and high dynamic range is proposed which can eliminate the interference of dense occlusions by measuring with almost continuous views, and simultaneously tackle the over/under exposure problems.
- Large-Scale Land Cover Classification in Gaofen-2 Satellite Imagery43
A novel land cover classification method is developed that can deal with large-scale data captured from widely distributed areas and different times and achieves outstanding classification accuracy compared with traditional methods.
- 42
- 41
- A Functional Representation for Graph Matching41
A functional representation for graph matching (FRGM) that aims to provide more geometric insights on the problem and reduce the space and time complexities and uses the representation of edge-attributes rather than the affinity matrix to reduced the space complexity and propose an efficient optimization strategy to reduce the time complexity.
- Anisotropic-Scale Junction Detection and Matching for Indoor Images41
A novel approach for junction detection and characterization is presented, which especially exploits the locally anisotropic geometries of a junction and estimates its scales by relying on an a-contrario model.
- Single Image Deraining With Continuous Rain Density Estimation40
This paper proposes a continuous density-guided network (CODE-Net) composed of a rain streak extractor and a denoiser, where the convolutional sparse coding (CSC) is exploited to filter out noises from the extracted rain streaks.
- Visual object tracking by correlation filters and online learning40
A feature integration object tracker named correlation filters and online learning (CFOL) is proposed, which estimates the target’s position and its corresponding correlation score using the same discriminative correlation filter with multi-features.
- 40
- 39
- Sat2Density: Faithful Density Learning from Satellite-Ground Image Pairs39
This paper aims to develop an accurate 3D geometry representation of satellite images using satellite-ground image pairs using a new approach, called Sat2Density, which utilizes the properties of ground-view panoramas for the sky and non-sky regions to learn faithful density fields of 3D scenes in a geometric perspective.
- 39
- Siamese networks with distractor-reduction method for long-term visual object tracking39
A hard negative mining method is proposed which can optimize the training process of the verification network and bridge the gap between the two sub-networks, and the semantic ambiguity in classification can be alleviated.
- Learning Regional Attraction for Line Segment Detection39
The problem of line segment detection (LSD) as a problem of region coloring is posed, and an end-to-end framework to learn attraction field maps for raw input images, followed by a squeeze module to detect line segments is developed.
- Learning Cross-View Visual Geo-Localization Without Ground Truth38
This study investigates the adaptation of frozen models for CVGL without requiring ground-truth pair labels, and proposes a self-supervised learning framework to train a learnable adapter for a frozen foundation model (FM) designed to map feature distributions from diverse views into a uniform space using unlabeled data exclusively.
- Generalizing Event-Based Motion Deblurring in Real-World Scenarios38
This work proposes a scale-aware network that allows flexible input spatial scales and enables learning from different temporal scales of motion blur, and develops a two-stage self-supervised learning scheme to fit real-world data distribution.
- Dynamic texture recognition by aggregating spatial and temporal features via ensemble SVMs38
This paper addresses the problem of dynamic texture recognition by aggregating spatial and temporal texture features via an ensemble SVM scheme, and bypassing the difficulties of simultaneously spatio-temporal description of DTs.
- OmniCity: Omnipotent City Understanding with Multi-Level and Multi-View Images37
OmniCity contains multi-view satellite images as well as street-level panorama and mono-view images, constituting over 100K pixel-wise annotated images that are well-aligned and collected from 25K geo-locations in New York City.
- 37
- 36
- 36
- Integration of synthetic aperture radar image segmentation method using Markov random field on region adjacency graph36
A modification method based on Gibbs sampler is proposed to correct edge errors, brought by the over-segmented algorithm, in the segmentations obtained by MRF-RAG.
- 35
- A Comparative Study of Sampling Analysis in the Scene Classification of Optical High-Spatial Resolution Remote Sensing Imagery35
The experimental results obtained on two commonly-used datasets using different feature learning methods show that random sampling can provide comparable and even better performance than all of the saliency-based strategies.
- 34
- Analysis of large-scale UAV images using a multi-scale hierarchical representation34
This work proposed a multi-scale hierarchical representation, i.e. binary partition tree, for analyzing large-scale UAV images, and merged the similar superpixels to build an object-based hierarchical structure by fully considering the spectral and spatial information of thesuperpixels and their topological relationships.
- NOPE-SAC: Neural One-Plane RANSAC for Sparse-View Planar 3D Reconstruction32
A novel Neural One-PlanE RANSAC framework (termed NOPE-SAC) that exerts excellent capability of neural networks to learn one-plane pose hypotheses from 3D plane correspondences that significantly improves the camera pose estimation for the two-view inputs with severe viewpoint changes.
- Learning to Super-Resolve Blurry Images With Events31
An event-enhanced degeneration model to consider the low spatial resolution, motion blurs, and event noises simultaneously is formulated and the proposed eSL-Net++ outperforms state-of-the-art methods by a large margin.
- 31
- 30
- 28
- 27
- AID++: An Updated Version of AID on Scene Classification27
This paper presents a larger-scale dataset, named as AID++, for aerial scene classification based on the AID dataset, which consists of more than 400,000 image samples that are semi-automatically annotated by using the existing the geographical data.
- 27
- 26
- Hidden Path Selection Network for Semantic Segmentation of Remote Sensing Images26
With the help of hidden variables deriving from an extra mini-branch, HPS-Net is able to tackle the inherent problem about inaccessible global optimums by adjusting the direct relationships between feature maps and pixelwise path selections in existing algorithms, which it is called hidden path selection.
- Retrieving Aerial Scene Images with Learned Deep Image-Sketch Features26
A framework to bridge the gap between sketches and aerial images is proposed and the Euclidean distance is used to measure the cross-domain similarity between aerial images and sketches.
- Fast Binary Coding for the Scene Classification of High-Resolution Remote Sensing Imagery26
A fast binary coding (FBC) method, inspired by the unsupervised feature learning technique and the binary feature descriptions, to effectively generate efficient discriminative scene representations of HRRS images is proposed.
- 26
- 26
- 26
- 25
- 25
- Static and Dynamic Texture Mixing Using Optimal Transport25
This paper tackles the problem of mixing static and dynamic texture by combining the statistical properties of an input set of images or videos by defining the distance between texture models, derive the geodesic path, and define the barycenter between several texture models.
- FreePoint: Unsupervised Point Cloud Instance Segmentation24
A novel framework, FreePoint, for underexplored unsupervised class-agnostic instance segmentation on point clouds, which performs significantly better than existing self-supervised pre-training methods with limited annotations and surpasses CSC by 6.0% in AP with 10% annotation masks.
- 3D Building Reconstruction from Monocular Remote Sensing Images with Multi-level Supervisions24
This work proposes MLS-BRN, a multi-level supervised building reconstruction network that can flexibly utilize training samples with different annotation levels to achieve better reconstruction results in an end-to-end manner and achieves competitive performance using much fewer 3D-annotated samples.
- Multi-object tracking with inter-feedback between detection and tracking24
This paper proposes a multi-type multi-object tracking algorithm, by introducing on-line inter-feedback information between the detection and tracking processes into the tracking-by-detection method and demonstrates that the proposed algorithm needs less detector and searching scale and can run in real time for many tracking applications.
- 23
- Zero-Assignment Constraint for Graph Matching With Outliers23
This work designs an efficient outlier-robust algorithm to significantly reduce the incorrect or redundant matchings caused by numerous outliers and provides elaborate theoretical analysis to the problem, i.e., GM with ZAC.
- Globally consistent correspondence of multiple feature sets using proximal Gauss–Seidel relaxation23
This paper addresses the problem of establishing a globally consistent correspondence among multiple (more than two) feature sets given the pairwise feature affinity information and proposes a novel optimization framework based on Proximal Gauss-Seidel Relaxation.
- 22
- On the Robustness of Object Detection Models on Aerial Images22
This article systematically evaluates the robustness of mainstream object detection models and finds that rotation-invariant modeling and enhanced backbone architectures can improve the robustness of models and increasing the capacity of Transformer-based backbones can strengthen their robustness.
- FuTH-Net: Fusing Temporal Relations and Holistic Features for Aerial Video Classification22
A novel deep neural network, termed FuTH-Net, is proposed, to model not only holistic features but also temporal relations for aerial video classification, and achieves the state-of-the-art results.
- Video object detection with a convolutional regression tracker22
This paper proposes to augment a well-trained image object detector with an efficient and effective class-agnostic convolutional regression tracker for the video object detection task, which improves the mean average precision (mAP) score for the image object detectors and the Seq-NMS post-processing.
- DiRS: On Creating Benchmark Datasets for Remote Sensing Image Interpretation22
This article discusses the problem of how to efficiently prepare a suitable benchmark dataset for RS image analysis, and presents some principles, i.e., diversity, richness, and scalability (called DiRS), on constructing benchmark datasets in efficient manners.
- Asymmetric Siamese Networks for Semantic Change Detection.22
An asymmetric siamese network (ASN) is presented to locate and identify semantic changes through feature pairs obtained from modules of widely different structures, which involve different spatial ranges and quantities of parameters to factor in the discrepancy across different land-cover distributions.
- GeoSay: A geometric saliency for extracting buildings in remote sensing images22
A geometric building index (GBI) is proposed for accurate building extraction, by computing the geometric saliency from VHSR-RS images, which preserves both the exact position and accurate shape of single buildings compared to existing methods.
- IMAGE STITCHING WITH PERSPECTIVE-PRESERVING WARPING22
A perspective-preserving warping for image stitching is suggested, which spatially combines local projective transformations and similarity transformation and thus the final warping can smoothly change from projective to similarity.
- Learning to See Through With Events21
This paper presents an Event-based SAI (E-SAI) method by relying on the asynchronous events with extremely low latency and high dynamic range acquired by an event camera to produce high-quality images from pure events.
- Deeply Unsupervised Patch Re-Identification for Pre-Training Object Detectors21
Deeply Unsupervised Patch Re-ID (DUPR), a simple yet effective method for unsupervised visual representation learning, outperforms state-of-the-art unsupervised pre-trainings and even the ImageNet supervised pre-training on various downstream tasks related to object detection.
- 21
- 21
- 21
- 19
- Detecting Line Segments in Motion-Blurred Images With Events19
This work designs a general frame-event feature fusion network to extract and fuse the detailed image textures and low-latency event edges and utilizes the state-of-the-art wireframe parsing networks to detect line segments on the fused feature map.
- Conditional Generative ConvNets for Exemplar-Based Texture Synthesis19
A conditional generative ConvNet (cgCNN) model which combines deep statistics and the probabilistic framework of generative convolutional neural networks (ConvNets) model, which can synthesize high quality dynamic, sound and image textures in a unified manner is presented.
- 19
- Active learning for training sample selection in remote sensing image classification using spatial information19
An AL approach with two-stage spatial computation (AL-2SC) is proposed to improve the selection of training samples and utilization of spatial information is very important for the improvement of AL performance.
- 19
- 19
- Anchor-based Robust Finetuning of Vision-Language Models18
This work proposes to compensate for the finetune process using auxiliary supervision with rich semantic information, which acts as anchors to preserve the OOD generalization in CLIP, thereby preserving the OOD generalization capabilities.
- Accurate Polygonal Mapping of Buildings in Satellite Imagery18
This paper addressed the issue of mask reversibility by exploiting the hierarchical supervision of bottom-level vertices, mid-level line segments and the high-level regional masks and proposed a novel interaction mechanism of feature embedding sourced from different levels of supervision signals to obtain reversible building masks for polygonal mapping of buildings.
- 18
- Implicit Euler ODE Networks for Single-Image Dehazing18
This paper proposes an efficient end-to-end multi-level implicit network (MI-Net) for the single image dehazing problem, and outperforms existing methods and achieves the state-of-the-art performance.
- 17
- NEAT: Distilling 3D Wireframes from Neural Attraction Fields17
The proposed NEAT enjoys the joint optimization of the neural fields and the global junctions from scratch, using view-dependent 2D observations without precomputed cross-view feature matching, demonstrating its superiority over state-of-the-art alternatives for 3D wireframe reconstruction.
- CAT: a coarse-to-fine attention tree for semantic change detection17
This paper tackles the SCD and LCM tasks simultaneously by proposing a coarse-to-fine attention tree (CAT) model, which outperforms state-of-the-art models and equip a label distribution learning loss upon the model to address label ambiguity in SCD/LCM.
- 17
- HoW-3D: Holistic 3D Wireframe Perception from a Single Image17
This paper presents a novel Deep Spatial Gestalt (DSG) model, which outperforms the previous wire-frame detectors in detecting the invisible line geometry in single-view images and is even very competitive with prior arts that take high-fidelity PointCloud as inputs on reconstructing 3D wireframes.
- 17
- Recent Advances and Opportunities in Scene Classification of Aerial Images with Deep Models17
This work deeply discusses the existing problems in scene classification task, and presents three open directions which it believes will be instructive for the researchers in this field.
- 17
- BIDIRECTIONAL MULTI-SCALE ATTENTION NETWORKS FOR SEMANTIC SEGMENTATION OF OBLIQUE UAV IMAGERY16
The novel bidirectional multi-scale attention networks, which fuse features from multiple scales bidirectionally for more adaptive and effective feature extraction are proposed.
- Robust visible-infrared image matching by exploiting dominant edge orientations16
The experimental results demonstrate the effectiveness of the proposed rotation invariant descriptor for matching visible and long wave infrared images as compared with state-of-the-art descriptors.
- Information processing for unmanned aerial vehicles (UAVs) in surveying, mapping, and navigation16
This special issue of Geo-spatial Information Science includes seven papers that cover topics mainly on the navigation, image analysis, and surveying/mapping applications of UAVs, and three papers concentrate on the analysis of the images taken by Uavs, which are usually of great volume and suffer from degradations.
- The UAVid Dataset for Video Semantic Segmentation16
This paper introduces a new high resolution UAV video semantic segmentation dataset as complement, UAVid, consisting of 30 video sequences capturing high resolution images and provides several deep learning baseline methods, among which the proposed novel Multi-Scale-Dilation net performs the best via multi-scale feature extraction.
- 16
- 15
- 15
- 15
- 15
- 14
- DAM-Net: Global Flood Detection from SAR Imagery Using Differential Attention Metric-Based Vision Transformers14
The proposed novel differential attention metric-based network (DAM-Net) comprises a weight-sharing Siamese backbone to obtain multi-scale change features of multi-temporal images and tokens containing high-level semantic information of water-body changes, and a temporal differential fusion module that integrates semantic tokens and change features to generate flood maps with reduced speckle noise.
- An Elliptic Centerness for Object Instance Segmentation in Aerial Images14
This work proposes an elliptic centerness (EC) for instance segmentation in aerial images, which can assign the proper centerness values to the intricate aerial instances and thus mitigate the performance degradation.
- Accurate Bridge Detection in Aerial Images With an Auxiliary Waterbody Extraction Task14
This article proposes an oriented bridge detection model with an auxiliary task of waterbody segmentation, which performs as guidance for bridge localization and proposes a background consistent instance switching method for online data augmentation to further improve the robustness of bridge detection.
- Texture Characterization Using Shape Co-Occurrence Patterns14
This paper presents a flexible shape-based texture representation using shape co-occurrence patterns of explicit shapes, which provides a flexible way to consider shape relationships and to compute high-order statistics on the tree.
- UNIC-Adapter: Unified Image-Instruction Adapter with Multi-Modal Transformer for Image Generation13
This paper proposes the unified image-instruction adapter (UNIC-Adapter) built on the Multi-Modal-Diffusion Transformer architecture, to enable flexible and controllable generation across diverse conditions without the need for multiple specialized models.
- Unsupervised Multiview UAV Image Geolocalization via Iterative Rendering13
This work proposes an unsupervised solution that lifts the scene representation to 3-D space from UAV observations for satellite image generation, providing a robust representation against view distortion and enabling generic CVGL for UAV images without feature fine-tuning or data-driven training.
- CrossZoom: Simultaneous Motion Deblurring and Event Super-Resolving13
This paper introduces CrossZoom, a novel unified neural Network (CZ-Net) to jointly recover sharp latent sequences within the exposure period of a blurry input and the corresponding High-Resolution (HR) events and presents a multi-scale blur-event fusion architecture that leverages the scale-variant properties and effectively fuses cross-modal information to achieve cross-enhancement.
- 13
- Cross-Level Multi-Instance Distillation for Self-Supervised Fine-Grained Visual Categorization12
The key idea is to consider the importance of each image patch in determining the fine-grained representation by multiple instance learning, which outperforms the contemporary methods by up to 10.14% and existing state-of-the-art self-supervised learning approaches by up to 19.78% on both top-1 accuracy and Rank-1 retrieval metric.
- 12
- 12
- Learning Fine-grained Domain Generalization via Hyperbolic State Space Hallucination12
A novel Hyperbolic State Space Hallucination (HSSH) method, which enriches the style diversity for the state embeddings by firstly extrapolating and then hallucinating the source images, and allows a better discernment of the fine-grained patterns.
- Extracting Building Footprints in SAR Images via Distilling Boundary Information From Optical Images12
- 11
- 11
- 11
- 11
- 11
- 11
- Oriented Tiny Object Detection: A Dataset, Benchmark, and Dynamic Unbiased Learning10
This study systematically introduces a new dataset, a benchmark, and a dynamic coarse-to-fine learning scheme, called DCFL, which achieves state-of-the-art accuracy, high efficiency, and remarkable versatility in object detection.
- HiCD: Change Detection in Quality-Varied Images via Hierarchical Correlation Distillation10
A hierarchical correlation distillation approach that compels the student model to replicate the correlations inherent in the teacher model, rather than focusing solely on individual features, ensures effective knowledge transfer while maintaining the student model’s training flexibility.
- 10
- 10
- 10
- Multi-Scale Weighted Branch Network for Remote Sensing Image Classification10
The proposed Multi-Scale Weighted Branch Network (MSWBN) outperforms extisting best models on the large scale remote sensing image classification dataset (GID) in terms of both efficiency and accuracy.
- Learning the Synthesizability of Dynamic Texture Samples10
This paper first defines the synthesizability of DT samples and characterize them by a set of spatiotemporal features, and proposes associating dynamic texture samples with synthesizable scores by learning regression models on a compiled dynamic texture dataset annotated in terms of synthesIZability.
- A generic framework for the structured abstraction of images10
Results show that it is possible to create abstract images evoking Malevitchs Suprematist school, while remaining grounded in the structure of digital images, by replacing all the shapes in the tree by simple geometric shapes.
- 10
- 10
- Optimal Transport Mixing of Gaussian Texture Models10
This paper focuses on stationary Gaussian texture models and derives the barycenter and geodesic path between models according to optimal transport, which allows the user to navigate inside the set of texture models, and perform texture synthesis from the obtained interpolated models.
- Semantic labelling of SAR images with conditional random fields on region adjacency graph10
Compared to the Markov random field (MRF) model on RAG, the proposed CRF-RAG model is more efficient to incorporate different measures of SAR images, such as scattering intensity, texture and image context, into a unified model.
- 9
- 9
- Self-Supervised Scene Dynamic Recovery from Rolling Shutter Images and Events9
An event-based SDR network within a self-supervised learning paradigm and an Event-based Inter/intra-frame Compensator (E-IC) is proposed to predict the per-pixel dynamic between arbitrary time intervals, including the temporal transition and spatial translation.
- Depth and DOF Cues Make A Better Defocus Blur Detector9
Inspired by the law of depth, depth of field (DOF), and defocus, an approach called D-DFFNet is proposed, which incorporates depth and DOF cues in an implicit manner, which allows the model to understand the defocus phenomenon in a more natural way.
- 9
- Adaptively Transforming Graph Matching9
This paper introduces an adaptively transforming graph matching (ATGM) method from the perspective of functional representation and demonstrates that the proposed method outperforms the state-of-the-art graph matching algorithms.
- 9
- 9
- Seeing Through Satellite Images at Street Views8
A novel approach is presented, Sat2Density++, to accomplish the goal of photo-realistic street-view panorama rendering by modeling these street-view specific elements in neural networks and it is demonstrated that this method is capable of rendering photorealistic street-view panoramas that are consistent across multiple views and faithful to the satellite image.
- Few-Shot Learning for Annotation-Efficient Nucleus Instance Segmentation8
This work adopts the meta-learning based FSL paradigm, and extends the basic definition of few-shot instance segmentation (FSIS) to generalized few-shot instance segmentation (GFSIS), introducing a structural guidance mechanism into the GFSIS network to cope with the intrinsic challenges of nucleus segmentation.
- Towards Generic and Controllable Attacks Against Object Detection8
A generic white-box attack, LGP, to blind mainstream object detectors with controllable perturbations and establishes an object-wise constraint that exploits foreground-background separation adaptively to induce the attachment of perturbation to foregrounds to fool the crucial components of ODs.
- 8
- 7
- 7
- Bypass Back-propagation: Optimization-based Structural Pruning for Large Language Models via Policy Gradient7
A novel optimization-based structural pruning that learns the pruning masks in a probabilistic space directly by optimizing the loss of the pruned model by decouple the Bernoulli parameters from LLM loss, facilitating efficient optimization via policy gradient estimator without back-propagation.
- QuadricsNet: Learning Concise Representation for Geometric Primitives in Point Clouds7
A novel framework to learn a concise geometric primitive representation for 3D point clouds using quadrics to represent diverse primitives with only 10 parameters and proposes the first end-to-end learning-based framework, namely QuadricsNet, to parse quadrics in point clouds.
- Application of deep learning in isolated tooth identification7
A deep learning model is developed that incorporates the state-of-the-art feature extractor and attention mechanism to classify isolated tooth based on a set of 6 photographs captured from multiple angles and has the potential to become a valuable tool for dental education and various applications in accurately identifying isolated tooth.
- MCVO: A Generic Visual Odometry for Arbitrarily Arranged Multi-Cameras7
This work proposes a robust visual odometry system for rigidly-bundled arbitrarily-arranged multi-cameras, namely MCVO, which can achieve metric-scale state estimation with high flexibility in the cameras' arrangement and obtains higher pose accuracy with better generalization ability.
- 7
- Robust Tiny Object Detection in Aerial Images amidst Label Noise7
A DeNoising Tiny Object Detector (DN-TOD), which incorporates a Class-aware Label Correction (CLC) scheme to address class shifts and a Trend-guided Learning Strategy (TLS) to handle bounding box noise to address vulnerability of object detectors to class shifts and inaccurate bounding boxes for tiny objects.
- SHIP DETECTION IN COSMO-SKYMED SAR IMAGERY USING A NOVEL CNN-BASED DETECTOR: A CASE STUDY FROM THE SUEZ CANAL7
A CNN-based attention-guided self-learning framework for ship detection from 3m high-resolution COSMO-SkyMed SAR imagery acquired in April 2021 via the Egyptian Suez Canal is proposed and outperforms current state-of-the-art detection models in terms of detection accuracy.
- 7
- Accurate Building Detection in VHR Remote Sensing Images Using Geometric Saliency7
This paper proposes a new geometric building index (GBI) for accurate building detection, which relies on the geometric saliency of building structures, and achieves very promising performance and meanwhile shows impressive generalization capability.
- Understanding Data Influence With Differential Approximation6
A new formulation to approximate a sample’s influence by accumulating the differences in influence between consecutive learning steps, which is term Diff-In and achieves significantly lower approximation error compared to existing influence estimators.
- Phase transitions in large language model compression6
Structural, numerical, and algebraic redundancy across pruning, quantization, and low-rank decomposition techniques are analyzed, enabling a criticality-aware compression framework that achieves near-lossless compression to 10% of the original size.
- 6
- Mitigating the Impact of Prominent Position Shift in Drone-based RGBT Object Detection6
This paper proposes to cast the cross-modality box shift issue as the label noise problem and address it on the fly via a novel Mean Teacher-based Cross-modality Box Correction head ensemble (CBC) via a novel Mean Teacher-based Cross-modality Box Correction head ensemble (CBC).
- Mitigating representation bias for class-incremental semantic segmentation of remote sensing images6
MiR replaces the traditional feature-classifier mode with the feature-SegToken interaction, leveraging implicit coarse-to-fine segmentation to mitigate representation bias, and a weighting strategy is introduced to adaptively adjust cross-entropy and distillation losses, effectively tackling issues of interstep imbalance and interstep similarity.
- Partial Wasserstein Adversarial Network for Non-rigid Point Set Registration6
A partial Wasserstein adversarial network (PWAN) is proposed, which is able to approximate the PW discrepancy by a neural network, and minimize it by gradient descent, and incorporates an efficient coherence regularizer for non-rigid transformations to avoid unrealistic deformations.
- 6
- Small object detection in forward-looking infrared images with sea clutter using context-driven Bayesian saliency model6
A context-driven Bayesian saliency model is presented to deal with detection ambiguity and scale variance in forward-looking infrared images with sea clutter and can be potentially applied to other circumstances where context is available for facilitating object detection.
- 6
- 6
- Sat3DGen: Comprehensive Street-Level 3D Scene Generation from Single Satellite Image5
This methodology enhances the feed-forward paradigm by integrating novel geometric constraints with a perspective-view training strategy, explicitly countering the primary sources of geometric error, and yields a dramatic leap in both 3D accuracy and photorealism.
- Self-supervised Shutter Unrolling with Events5
An event-based CGVR network within a self-supervised learning paradigm is proposed, and an Event-based Inter/intra-frame Compensator (E-IC) is proposed to predict the per-pixel dynamic between arbitrary time intervals, including the temporal transition and spatial translation.
- Fuse Before Transfer: Knowledge Fusion for Heterogeneous Distillation5
This fusion combines the advantages of both cross-architecture inductive biases and module functions by merging different combinations of convolution, attention, and MLP modules derived directly from student and teacher module functions.
- Rethinking Scale Imbalance in Semi-supervised Object Detection for Aerial Images5
A novel Scale-discriminative Semi-Supervised Object Detection (S^3OD) learning pipeline for aerial images that combines Size-aware Adaptive Thresholding (SAT), Size-rebalanced Label Assignment (SLA), and Teacher-guided Negative Learning (TNL) are proposed to warrant scale unbiased learning.
- 5
- 5
- 5
- 5
- An SRTM-Aided Epipolar Resampling Method for Multi-Source High-Resolution Satellite Stereo Observation5
A novel SRTM (Shuttle Radar Topography Mission)-aided approach is proposed, where a point is transformed from the original image-space to the epipolar image- space through a global rotation, followed by a block-wise homography transformation.
- Image Stitching Using Smoothly Planar Homography5
A smoothly planar homography model for image stitching is proposed, by considering the multi-plane geometry of natural scene, and a novel alignment-guided seam composition to handle parallax is introduced.
- 5
- 5
- 5
- 5
- 1 Semantic Labeling of SAR Images with Hierarchical Markov Aspect Models5
A new approach for semantic labeling of SAR imagery is proposed based on hier- archical Markov aspect model (HMAM) with weak supervision, which can improve labeling performance significantly with only a modest increase in learning and inference complexity than aspect model.
- 5
- Unifying UAV Cross-View Geo-Localization via 3D Geometric Perception4
A geometry-aware UAV geo-localization framework that explicitly models the 3D scene geometry to unify coarse place recognition and fine-grained pose estimation within a single inference pipeline is proposed, achieving robust meter-level localization accuracy and improved generalization in complex urban environments.
- Rejoining fragmented ancient bamboo slips with physics-driven deep learning4
W WisePanda provides a new paradigm for addressing data scarcity in ancient artifact restoration through physics-driven machine learning and yields a substantial and statistically significant increase in matching accuracy, accelerating archeologists’ efficiency in rejoining fragmented bamboo slips.
- DragOSM: Extract Building Roofs and Footprints from Aerial Images by Aligning Historical Labels4
Drag OpenStreetMap Labels (DragOSM), a novel model designed to align dislocated historical labels with roofs and footprints, is proposed, which formulates the label alignment as an interactive denoising process, modeling the positional discrepancy as a Gaussian distribution.
- 4
- Exploring Scene Affinity for Semi-Supervised LiDAR Semantic Segmentation4
This paper explores scene affinity (AIScene), namely intra-scene consistency and inter-scene correlation, for semi-supervised LiDAR semantic segmentation in driving scenes, adopting teacher-student training and improving the semi-supervised performance of segmentation models.
- 4
- GOOD: Towards Domain Generalized Orientated Object Detection4
The task of domain generalized oriented object detection is proposed, which intends to explore the generalization of oriented object detectors on arbitrary unseen target domains to address challenges of cross-domain style variation.
- 4
- PDCA-FORMER: PRIOR-DIAGONAL CROSS ATTENTION-GUIDED TRANSFORMER FOR FLOOD MAPPING FROM SAR IMAGERY: A CASE IN KHARTOUM4
A novel Prior-Diagonal Cross Attention-guided transformer (PDCA-Former) network for flood mapping from SAR images that can be quickly generalized to other regions and significantly aid disaster management efforts on vulnerable communities is proposed.
- 4
- 4
- LIP: Learning Instance Propagation for Video Object Segmentation4
This paper proposes a single end-to-end trainable deep neural network, convolutional gated recurrent Mask-RCNN, for tackling the semi-supervised VOS task, which takes advantage of both the instance segmentation network (Mask- RCNN) and the visual memory module (Conv-GRU).
- 4
- 4
- Texture Segmentation by Grouping Ellipse Ensembles via Active Contours4
This work proposes to represent a texture image through a tree of ellipse-based features, which are derived from the level lines of the image, and achieves a segmentation of the texture image by grouping all the ellipses into several subsets according to some statistical measurements.
- 3
- UniCalib: Targetless LiDAR-camera Calibration via Probabilistic Flow on Unified Depth Representations3
UniCalib is a novel method that performs calibration by estimating a probabilistic flow on unified depth representations by converting both the camera images and the sparse LiDAR points into unified, dense depth maps, enabling a unified encoder to learn consistent features.
- MetaDent: Labeling Clinical Images for Vision-Language Models in Dentistry3
MetaDent presents MetaDent, a comprehensive resource that includes a novel and large-scale dentistry image dataset collected from clinical, public, and web sources; a semistructured annotation framework designed to capture the hierarchical and clinically nuanced nature of dental photography; and comprehensive benchmark suites for evaluating state-of-the-art VLMs on clinical image understanding.
- 3
- 3
- 3
- Partial Distribution Matching via Partial Wasserstein Adversarial Networks3
The Kantorovich-Rubinstein duality for the partial Wasserstein-1 (PW) discrepancy is theoretically derived, and a partial Wasserstein adversarial network (PWAN) is developed that efficiently approximates the PW discrepancy based on this dual form.
- DMTG: One-Shot Differentiable Multi-Task Grouping3
This paper proposes to simultaneously identify the best task groups from 2^N candidates and train the model weights simultaneously in one-shot, with the high-order task-affinity fully exploited, to address Multi-Task Learning with a large number of tasks by Multi-Task Grouping.
- 3
- 3
- 3
- Editorial3
Os principais conteúdos discutidos e problematizados nos textos versam em torno oficio de temas como formação de professores, ensino remoto emergencial, manuais didáticos, sala of aula invertida, multiletramentos, letramento crítico e midiático.
- 3
- 3
- 3
- 3
- 3
- Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random Fields3
A fast, precise, and highly scalable semantic segmentation algorithm that incorporates several kinds of local appearance features, example-based spatial layout priors, and neighborhood-level and global contextual information, leading to a scalable algorithm that is both faster and more accurate than existing patch-level approaches.
- Pixel-level SAR image fusion based on turbo iterative3
A pixel-level image fusion method based on turbo iterative for Synthetic Aperture Radar (SAR) and other sensor images is proposed, and the fusion results for SAR images show that the proposed method has a good performance.
- Generalized Small Object Detection:A Point-Prompted Paradigm and Benchmark2
DEAL (DEtect Any smalL object), a scalable and transferable point-prompted detection framework that learns robust, prompt-conditioned representations from large-scale data, is developed.
- Change-aware multi-temporal cloud removal2
A change-aware multi-temporal cloud removal method, CA-MTCR, which utilizes SAR images, capable of penetrating clouds to capture terrain information beneath, to detect changes and optimize the usage of multi-temporal data is proposed.
- Model Hemorrhage and the Robustness Limits of Large Language Models2
This work establishes foundational metrics for evaluating model stability during adaptation, providing practical guidelines for maintaining performance while enabling efficient LLM deployment and advancing understanding of neural network resilience under architectural transformations.
- 2
- High-Fidelity 3D Imaging of Dental Scenes Using Gaussian Splatting2
An advanced novel view synthesis (NVS) method called Gaussian splatting (GS), a differentiable image-based rendering approach, is explored to assess its feasibility for dental scene capturing, showing great potential for developing a convenient and cost-effective method of capturing dental scenes.
- 2
- 2
- High-Quality Pseudo-Labeling for Point Cloud Segmentation With Scene-Level Annotation2
A high-quality pseudo-label generation framework by exploring contemporary multi-modal information and region-point semantic consistency and a region-point semantic consistency module that produces regional semantics through a region-voting strategy derived from point-level semantics is proposed.
- 2
- Intensity Field Decomposition for Tissue-Guided Neural Tomography2
A novel sparse-view CBCT reconstruction method, which empowers the neural field with human tissue regularization, and achieves comparable reconstruction quality with fewer projections and faster convergence compared to state-of-the-art neural rendering based methods.
- QuadricsReg: Large-Scale Point Cloud Registration using Quadric Primitives2
A novel point cloud registration method, i.e., QuadricsReg, is introduced, which leverages concise quadrics primitives to represent scenes and utilizes their geometric characteristics to establish correspondences for 6-DoF transformation estimation.
- Aux-NAS: Exploiting Auxiliary Labels with Negligibly Extra Inference Cost2
This work proposes a novel method with evolving networks where only primary-to-auxiliary links exist as the cross-task connections after convergence, which can be removed during the primary task inference, resulting in a single-task inference cost.
- Dual Relation Mining Network for Zero-Shot Learning2
This work introduces a Dual Attention Block (DAB) for visual-semantic relationship mining, which enriches visual information by multi-level feature fusion and conducts spatial attention for visual to semantic embedding and an attribute-guided channel attention is utilized to decouple entangled semantic features.
- Density-aware Object Detection in Aerial Images2
A DeDet is designed to enhance the detector's performance in detecting densely arranged objects by conquering the heuristic that the sample assignment and feature extraction are performed independently for each object.
- Recovering Continuous Scene Dynamics from A Single Blurry Image with Events2
An Implicit Video Function is learned to represent a single motion blurred image with concurrent events, enabling the latent sharp image restoration of arbitrary timestamps in the range of imaging exposures, and a dual attention transformer is proposed to efficiently leverage merits from both modalities.
- JBC: Joint Boost Clustering method for synthesis aperture radar images2
A clustering method based on Joint Boost for Synthesis Aperture Radar images is proposed, but substitute weak learns with basic clustering algorithm, which shows that the proposed method has a good performance.
- Interacted Planes Reveal 3D Line Mapping1
LiP-Map pioneers the integration of planar topology into 3D line mapping, not by imposing pairwise coplanarity constraints but by explicitly constructing interactions between plane and line primitives, thus offering a principled route toward structured reconstruction in man-made environments.
- LaVIDE: Language-Prompted Satellite Change Detection via Map-Image Alignment1
This work proposes a novel framework, Language-VIsion Discriminator for dEtecting changes, LaVIDE, which bridges the semantic gap between high-level map categories and low-level image details using language as an intermediary, and introduces restricted prompt learning to generate context-aware textual prompts that align map semantics with image content.
- 1
- Continual Vision-Language Learning for Remote Sensing: Benchmarking and Analysis1
CLeaRS is presented, a comprehensive benchmark for continual vision-language learning in remote sensing that comprises 10 curated subsets with over 207k image-text pairs, spanning diverse interpretation tasks, sensing modalities, and application scenarios, to systematically assess continual adaptation.
- EVDI++: Event-Based Video Deblurring and Interpolation via Self-Supervised Learning1
EvDI++ is introduced, a unified self-supervised framework for Event-based Video Deblurring and Interpolation that leverages the high temporal resolution of event cameras to mitigate motion blur and enable intermediate frame prediction.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Towards Human-Level 3D Relative Pose Estimation: Generalizable, Training-Free, With Single Reference1
A novel 3D generalizable relative pose estimation method by elaborating 3D/2.5D shape perception with a 2.5D shape from an RGB-D reference, fulfilling the render-and-compare paradigm with an off-the-shelf differentiable renderer, and leveraging the semantic cues from a pretrained model like DINOv2.
- Efficient Rural Building Segmentation via a Multilevel Decoding Network1
A multilevel decoding network model is presented that classifies the input image on the patch and image levels according to the distribution of buildings according to the distribution of buildings, which reduces the amount of inference time.
- LUAI Challenge 2021 on Learning to Understand Aerial Images1
The results of Learning to Understand Aerial Images (LUAI) 2021 challenge held on ICCV’2021, which focuses on object detection and semantic segmentation in aerial images, receive a total of 146 registrations on the three tasks.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- Spatio-Temporal Context of Mid-level Features for Activity Recognition1
A framework for both computing mid-level features and taking into account the context information between them is suggested and a spatio-temporal context kernel (STCK) is introduced, which not only takes into consideration the local properties of features but also considers their spatial and temporal context information.
- Some Geometric Methods for the Analysis of Images and Textures1
A new method for texture analysis that in spirit is similar to morphological granulometries, while allowing a high degree of geometrical and radiometric invariances and a general approach for the abstraction of images, the aim of which is to automatically generate abstract images from realistic photographs.
- –
- –
- –
- –
- –
- Endpoint‐Aware Line Segment Matching With Geometry‐Based Prediction–
This paper proposes an endpoint‐aware line segment matching method with geometry‐based prediction (GPMatcher) to predict the projected positions of endpoint pairs of line fragments and directly find the most matched line.
- –
- Timestep-Aware Block Masking for Efficient Diffusion Model Inference–
Experimental results show that by treating the denoising process as a sequence of optimized computational paths, the proposed novel framework achieves a superior balance between sampling speed and generative quality.
- –
- Hierarchical Fine-Grained Aerial Object Detection–
ExpertDet, a scheme that incorporates expert-informed cues to enhance fine-grained aerial object detection, is presented and Vision-aware Masked Attribute Modeling (VMAM), which aligns attribute semantics with visual structures by reconstructing randomly masked attributes from visual cues, enabling the detector to capture subtle structural distinctions.
- –
- –
- –
- –
- Observe Less, Understand More: Cost-aware Cross-scale Observation for Remote Sensing Understanding–
This work forms cross-scale remote sensing understanding as a unified cost-aware problem that couples fine-grained HR sampling with cross-patch representation prediction, enabling more effective task reasoning with fewer HR observations.
- –
- Event-Guided Super-Resolving Blurry Image via Asymmetric Integral Driven Consistency–
The proposed SE-SRB introduces a self-supervised learning paradigm based on asymmetric integral driven consistency, which enforces temporal coherence between predictions derived from RGB and asynchronous event streams at different time points.
- –
- UHR-DETR: Efficient End-to-End Small Object Detection for Ultrahigh-Resolution Remote Sensing Imagery–
This work proposes UHR-DETR, an efficient end-to-end transformer-based detector designed for UHR imagery, and introduces a coverage-maximizing sparse encoder that dynamically allocates finite computational resources to informative high-resolution regions, ensuring maximum object coverage with minimal spatial redundancy.
- –
- –
- –
- –
- –
- Mask Clustering-Based Annotation Engine for Large-Scale Submeter Land Cover Mapping–
The mask clustering-based annotation engine (MCAE), which treats semantically consistent mask groups as the minimal annotating units to enable efficient, simultaneous annotation of multiple instances, significantly improves annotation efficiency by one to two orders of magnitude, while preserving label quality, semantic diversity, and spatial representativeness
- –
- –
- –
- –
- 2D-3D Pose Tracking with Multi-View Constraints–
This work proposes a new 2D-3D pose tracking framework, which consists of a front-end hybrid flow estimation network for consecutive frames and a back-end pose optimization module, and designs a cross-modal consistency-based loss to incorporate the multi-view constraints during the training and inference process.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Multi-level max-margin analysis for semantic classification of satellite images–
The categorization performances on 21-class land-use dataset have demonstrated that the proposed model in multi-level max-margin analysis for semantic classification for high-resolution satellite images can distinguish different categories of land- use scenes reasonably.
- –
- –
- –
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.