GaussianFusion presents a 3D Gaussian-based framework that unifies multi-modal features in continuous space for 3D object detection and semantic occupancy, reporting gains over BEVFusion and GaussFormer on nuScenes.
hub Mixed citations
Deformable DETR: Deformable Transformers for End-to-End Object Detection
Mixed citation behavior. Most common role is background (67%).
91
Pith papers citing it
Background
67% of classified citations
abstract
DETR has been recently proposed to eliminate the need for many hand-designed components in object detection while demonstrating good performance. However, it suffers from slow convergence and limited feature spatial resolution, due to the limitation of Transformer attention modules in processing image feature maps. To mitigate these issues, we proposed Deformable DETR, whose attention modules only attend to a small set of key sampling points around a reference. Deformable DETR can achieve better performance than DETR (especially on small objects) with 10 times less training epochs. Extensive experiments on the COCO benchmark demonstrate the effectiveness of our approach. Code is released at https://github.com/fundamentalvision/Deformable-DETR.
hub tools
citation-role summary
background 6
method 2
baseline 1
citation-polarity summary
claims ledger
- abstract DETR has been recently proposed to eliminate the need for many hand-designed components in object detection while demonstrating good performance. However, it suffers from slow convergence and limited feature spatial resolution, due to the limitation of Transformer attention modules in processing image feature maps. To mitigate these issues, we proposed Deformable DETR, whose attention modules only attend to a small set of key sampling points around a reference. Deformable DETR can achieve better performance than DETR (especially on small objects) with 10 times less training epochs. Extensive e
co-cited works
representative citing papers
MVDGC unifies BEV and image-view pedestrian localization into one task via 3D cylindrical queries that enforce dual geometric constraints between views.
A fusion module for satellite imagery and planimetric maps reduces mean localization error by 30.13% over single-modality state-of-the-art methods in cross-view tasks.
FlowOVD applies rectified flow to generate continuous latent query dynamics for text-conditioned open-vocabulary detection, reporting 49.5 AP on COCO and 31.5 AP on LVIS.
VisHarness learns a reinforcement-learned policy to harness specialized visual experts via multi-turn interactions and dynamic visual memory archiving, outperforming general models on four visual reasoning benchmarks.
Presents the first multispectral dataset for fine-grained small-UAV detection and a dual-stream MFDNet baseline that gains 6.2% AP50 over RGB-only detectors by using spectral material cues.
UniTopo unifies lane detection and topology reasoning into a single perception model, outperforming prior methods on OpenLane-V2 benchmarks with TOP_ll scores of 30.1% and 31.8%.
InterMesh explicitly incorporates human-object interaction semantics into multi-person mesh recovery via a detector and two lightweight modules, delivering up to 9.9% MPJPE reduction on interaction-heavy datasets.
A YOLO26 model trained on four leaf segmentation datasets reaches 83.9% mean mAP50-95 on their test sets but only 40.2% on a new 23-species benchmark, revealing substantial cross-domain generalization gaps.
ConFusion reaches 59.1 mAP and 65.6 NDS on nuScenes validation by combining heterogeneous queries with QMix cross-attention and QSwap feature exchange.
CoEvoer is a new cross-dependency transformer framework for upper-body expressive human pose and shape estimation that achieves state-of-the-art performance by enabling mutual enhancement between body parts.
HELP uses heatmap-guided positional embeddings and a gradient mask to suppress background noise in queries, enabling efficient small-object detection with fewer decoder layers and parameters.
Topology-preserving synthetic P&IDs generated by seeding from real drawings enable models trained solely on synthetics to achieve 63.8% edge mAP on real P&ID benchmarks, closing most of the gap to real-data training.
The work introduces the ORVOS task, the ORVOSB benchmark with causal annotations across 210 videos, and a baseline using updated prompts plus a temporal token reservoir.
YUV20K is a complexity-driven VCOD benchmark with 24k annotated frames, paired with a model using Motion Feature Stabilization via semantic primitives and Trajectory-Aware Alignment via deformable sampling that outperforms prior methods.
DinoRADE reports a radar-centered multi-class detection pipeline that fuses dense radar tensors with DINOv3 features via deformable attention and outperforms prior radar-camera methods by 12.1% on the K-Radar dataset across weather conditions.
Bridge-STG decouples spatio-temporal alignment via semantic bridging and query-guided localization modules to achieve state-of-the-art m_vIoU of 34.3 on VidSTG among MLLM methods.
WUTDet is a 100K-image ship detection dataset with benchmarks indicating Transformer models outperform CNN and Mamba architectures in accuracy and small-object detection for complex maritime environments.
MoCA3D formulates monocular 3D box prediction as dense pixel-space tasks using corner heatmaps and depth maps, with a new PAG metric for image-plane evaluation.
SAM 3 introduces promptable concept segmentation that doubles accuracy of prior systems on images and videos while improving standard SAM segmentation performance.
A hierarchical spiking transformer using Q-K attention achieves 85.65% top-1 accuracy on ImageNet-1K, the first direct-trained SNN to exceed 85%.
RT-SFOD adapts dual-head detectors like YOLOv10 for source-free object detection via DHF pseudo-label fusion and MARD loss, delivering 1.4-3.5% mAP gains with 1.3x higher throughput and ~2x fewer parameters than prior SFOD methods.
DRVR uses range-view and geometry-aware voxel-view encoders plus fusion to deliver 5.4% higher mIoU and 2.1x faster inference than multi-sweep baselines on nuScenes-Occupancy from single sweeps.
Modular Diffusion Models decompose diffusion into task-specific modules to model distributions over structured visual outputs for detection, segmentation, and scene graph generation.
citing papers explorer
-
GaussianFusion: Unified 3D Gaussian Representation for Multi-Modal Fusion Perception
GaussianFusion presents a 3D Gaussian-based framework that unifies multi-modal features in continuous space for 3D object detection and semantic occupancy, reporting gains over BEVFusion and GaussFormer on nuScenes.
-
MVDGC: Joint 3D and 2D Multi-view Pedestrian Detection via Dual Geometric Constraints
MVDGC unifies BEV and image-view pedestrian localization into one task via 3D cylindrical queries that enforce dual geometric constraints between views.
-
Fusing Satellite Imagery and Planimetric Maps for Cross-View Localization
A fusion module for satellite imagery and planimetric maps reduces mean localization error by 30.13% over single-modality state-of-the-art methods in cross-view tasks.
-
FlowOVD: Learning Generative Latent Flows for Zero-shot Open-vocabulary Detection
FlowOVD applies rectified flow to generate continuous latent query dynamics for text-conditioned open-vocabulary detection, reporting 49.5 AP on COCO and 31.5 AP on LVIS.
-
Train the Agent, Not the Expert: Learning to Harness Heterogeneous Experts for Multi-Turn Visual Reasoning
VisHarness learns a reinforcement-learned policy to harness specialized visual experts via multi-turn interactions and dynamic visual memory archiving, outperforming general models on four visual reasoning benchmarks.
-
Towards UAV Detection in the Real World: A New Multispectral Dataset UAVNet-MS and a New Method
Presents the first multispectral dataset for fine-grained small-UAV detection and a dual-stream MFDNet baseline that gains 6.2% AP50 over RGB-only detectors by using spectral material cues.
-
Unified Modeling of Lane and Lane Topology for Driving Scene Reasoning
UniTopo unifies lane detection and topology reasoning into a single perception model, outperforming prior methods on OpenLane-V2 benchmarks with TOP_ll scores of 30.1% and 31.8%.
-
InterMesh: Explicit Interaction-Aware End-to-End Multi-Person Human Mesh Recovery
InterMesh explicitly incorporates human-object interaction semantics into multi-person mesh recovery via a detector and two lightweight modules, delivering up to 9.9% MPJPE reduction on interaction-heavy datasets.
-
ReLeaf: Benchmarking Leaf Segmentation across Domains and Species
A YOLO26 model trained on four leaf segmentation datasets reaches 83.9% mean mAP50-95 on their test sets but only 40.2% on a new 23-species benchmark, revealing substantial cross-domain generalization gaps.
-
Control Your Queries: Heterogeneous Query Interaction for Camera-Radar Fusion
ConFusion reaches 59.1 mAP and 65.6 NDS on nuScenes validation by combining heterogeneous queries with QMix cross-attention and QSwap feature exchange.
-
Chatting about Upper-Body Expressive Human Pose and Shape Estimation
CoEvoer is a new cross-dependency transformer framework for upper-body expressive human pose and shape estimation that achieves state-of-the-art performance by enabling mutual enhancement between body parts.
-
Learning Where to Embed: Noise-Aware Positional Embedding for Query Retrieval in Small-Object Detection
HELP uses heatmap-guided positional embeddings and a gradient mask to suppress background noise in queries, enabling efficient small-object detection with fewer decoder layers and parameters.
-
SynthPID: P&ID digitization from Topology-Preserving Synthetic Data
Topology-preserving synthetic P&IDs generated by seeding from real drawings enable models trained solely on synthetics to achieve 63.8% edge mAP on real P&ID benchmarks, closing most of the gap to real-data training.
-
Online Reasoning Video Object Segmentation
The work introduces the ORVOS task, the ORVOSB benchmark with causal annotations across 210 videos, and a baseline using updated prompts plus a temporal token reservoir.
-
YUV20K: A Complexity-Driven Benchmark and Trajectory-Aware Alignment Model for Video Camouflaged Object Detection
YUV20K is a complexity-driven VCOD benchmark with 24k annotated frames, paired with a model using Motion Feature Stabilization via semantic primitives and Trajectory-Aware Alignment via deformable sampling that outperforms prior methods.
-
DinoRADE: Full Spectral Radar-Camera Fusion with Vision Foundation Model Features for Multi-class Object Detection in Adverse Weather
DinoRADE reports a radar-centered multi-class detection pipeline that fuses dense radar tensors with DINOv3 features via deformable attention and outperforms prior radar-camera methods by 12.1% on the K-Radar dataset across weather conditions.
-
Bridging Time and Space: Decoupled Spatio-Temporal Alignment for Video Grounding
Bridge-STG decouples spatio-temporal alignment via semantic bridging and query-guided localization modules to achieve state-of-the-art m_vIoU of 34.3 on VidSTG among MLLM methods.
-
WUTDet: A 100K-Scale Ship Detection Dataset and Benchmarks with Dense Small Objects
WUTDet is a 100K-image ship detection dataset with benchmarks indicating Transformer models outperform CNN and Mamba architectures in accuracy and small-object detection for complex maritime environments.
-
MoCA3D: Monocular 3D Bounding Box Prediction in the Image Plane
MoCA3D formulates monocular 3D box prediction as dense pixel-space tasks using corner heatmaps and depth maps, with a new PAG metric for image-plane evaluation.
-
SAM 3: Segment Anything with Concepts
SAM 3 introduces promptable concept segmentation that doubles accuracy of prior systems on images and videos while improving standard SAM segmentation performance.
-
QKFormer: Hierarchical Spiking Transformer using Q-K Attention
A hierarchical spiking transformer using Q-K attention achieves 85.65% top-1 accuracy on ImageNet-1K, the first direct-trained SNN to exceed 85%.
-
Real-Time Source-Free Object Detection
RT-SFOD adapts dual-head detectors like YOLOv10 for source-free object detection via DHF pseudo-label fusion and MARD loss, delivering 1.4-3.5% mAP gains with 1.3x higher throughput and ~2x fewer parameters than prior SFOD methods.
-
Semantic Occupancy Prediction with Dual Range-Voxel Representation
DRVR uses range-view and geometry-aware voxel-view encoders plus fusion to deliver 5.4% higher mIoU and 2.1x faster inference than multi-sweep baselines on nuScenes-Occupancy from single sweeps.
-
Modular Diffusion Models for Structured Visual Recognition
Modular Diffusion Models decompose diffusion into task-specific modules to model distributions over structured visual outputs for detection, segmentation, and scene graph generation.
-
Multi-Modal Hyper-Graph Fusion for Low-Light Crowd Counting
Introduces LCNet using Multi-Modal Hyper-Graph Fusion and Deformable Rectangular Sparse Attention to achieve SOTA low-light crowd counting on three new benchmarks.
-
MambaCount: Efficient Text-guided Open-vocabulary Object Counting with Spatial Sparse State Space Duality Block
MambaCount uses S^4D blocks with spatial token selection and multi-granularity prototypes to reach MAE 12.23 on FSC-147 for open-vocabulary counting while keeping linear complexity.
-
Deformba: Vision State Space Model with Adaptive State Fusion
Deformba introduces context-adaptive state fusion to vision SSMs for better spatial augmentation and cross-stream interactions, showing strong results on 2D classification/detection/segmentation and 3D BEV perception benchmarks.
-
Vision Foundation Models as Generalist Tokenizers for Image Generation
VFMTok builds a generalist image tokenizer on frozen VFMs using adaptive quantization and semantic alignment, delivering gFID 1.36 for autoregressive and 1.25 for continuous generation on ImageNet with 3x faster convergence.
-
Invaria: Learning Scale and Density Invariance in Point Clouds via Next-Resolution Prediction
Invaria trains point cloud encoders with next-resolution prediction to learn scale and density invariant features, yielding higher mIoU on ScanNet under lower resolution and scaled objects while using a smaller model.
-
SToRe3D: Sparse Token Relevance in ViTs for Efficient Multi-View 3D Object Detection
SToRe3D delivers up to 3x faster inference for multi-view 3D object detection in ViTs by selecting relevant 2D tokens and 3D queries via mutual relevance heads with only marginal accuracy loss.
-
Deep Probabilistic Unfolding for Quantized Compressive Sensing
A probabilistic unfolding network with stable likelihood projection and dual-domain Mamba achieves state-of-the-art reconstruction in quantized compressive sensing.
-
Curvature-Aware Captioning:Leveraging Geodesic Attention for 3D Scene Understanding
A new framework combines self-attention on the Oblique manifold with bidirectional geodesic cross-attention on the Lorentz hyperboloid to improve both localization accuracy and descriptive coherence in 3D dense captioning.
-
A Novel Graph-Regulated Disentangling Mamba Model with Sparse Tokens for Enhanced Tree Species Classification from MODIS Time Series
A graph-regulated disentangling Mamba model with sparse tokens achieves 93.94% accuracy classifying tree species from MODIS time series in Alberta and outperforms twelve prior models.
-
Physical Adversarial Clothing Evades Visible-Thermal Detectors via Non-Overlapping RGB-T Pattern
Non-overlapping RGB-T adversarial patterns on clothing, optimized with spatial discrete-continuous optimization, achieve high attack success rates against multiple RGB-T detector fusion architectures in both digital and physical evaluations.
-
FUN: A Focal U-Net Combining Reconstruction and Object Detection for Snapshot Spectral Imaging
FUN is an end-to-end Focal U-Net that performs joint hyperspectral image reconstruction and object detection via multi-task learning with focal modulation, achieving SOTA results with 40% fewer parameters and a new 363-image dataset.
-
ViCrop-Det: Spatial Attention Entropy Guided Cropping for Training-Free Small-Object Detection
ViCrop-Det uses spatial attention entropy from the decoder to dynamically crop and refine small-object regions in transformer detectors during inference.
-
GateMOT: Q-Gated Attention for Dense Object Tracking
GateMOT proposes Q-Gated Attention to enable linear-complexity, spatially aware attention for state-of-the-art dense object tracking on benchmarks like BEE24.
-
OneDrive: Unified Multi-Paradigm Driving with Vision-Language-Action Models
OneDrive unifies heterogeneous decoding in a single VLM transformer decoder for end-to-end driving, achieving 0.28 L2 error and 0.18 collision rate on nuScenes plus 86.8 PDMS on NAVSIM.
-
Weakly-Supervised Referring Video Object Segmentation through Text Supervision
WSRVOS enables referring video object segmentation with text-only supervision by combining MLLM-based expression augmentation, multimodal feature interaction, pseudo-mask fusion, and temporal ranking constraints.
-
HiProto: Hierarchical Prototype Learning for Interpretable Object Detection Under Low-quality Conditions
HiProto uses hierarchical prototypes with RPC-Loss, PR-Loss, and SPLGS to deliver competitive, interpretable object detection on low-quality datasets like ExDark and RTTS.
-
Improving Layout Representation Learning Across Inconsistently Annotated Datasets via Agentic Harmonization
VLM-based harmonization of inconsistent annotations across two document layout corpora raises detection F-score from 0.860 to 0.883 and table TEDS from 0.750 to 0.814 while tightening embedding clusters.
-
Telescope: Learnable Hyperbolic Foveation for Ultra-Long-Range Object Detection
Telescope uses learnable hyperbolic foveation to deliver a 76% relative mAP gain (0.185 to 0.326) for objects beyond 250 meters while keeping overhead low.
-
Geometrical Cross-Attention and Nonvoid Voxelization for Efficient 3D Medical Image Segmentation
GCNV-Net achieves state-of-the-art accuracy on multiple 3D medical segmentation benchmarks while cutting FLOPs by 56% and inference latency by 68% through dynamic nonvoid voxelization and geometric attention.
-
PET-DINO: Unifying Visual Cues into Grounding DINO with Prompt-Enriched Training
PET-DINO unifies visual and text prompts in Grounding DINO via an alignment-friendly generation module and prompt-enriched training strategies to improve zero-shot open-set object detection.
-
Focus on What Really Matters in Low-Altitude Governance: A Management-Centric Multi-Modal Benchmark with Implicitly Coordinated Vision-Language Reasoning Framework
Presents the first management-oriented multi-modal benchmark GovLA-10K and a vision-language reasoning framework GovLA-Reasoner with a spatially-aware adapter for low-altitude aerial perception.
-
ShelfGaussian: Shelf-Supervised Open-Vocabulary Gaussian-based 3D Scene Understanding
ShelfGaussian achieves state-of-the-art zero-shot semantic occupancy prediction on Occ3D-nuScenes by jointly supervising Gaussian representations with vision foundation model features at 2D image and 3D scene levels.
-
YOLOv12: Attention-Centric Real-Time Object Detectors
YOLOv12 is a new attention-based real-time object detector that reports higher accuracy than YOLOv10, YOLOv11, and RT-DETR variants at comparable or better speed and efficiency.
-
TimberVision: A Multi-Task Dataset and Framework for Log-Component Segmentation and Tracking in Autonomous Forestry Operations
Introduces TimberVision dataset and multi-task framework for log-component segmentation, detection, and tracking in forestry operations using RGB images.
-
Sa2VA: Marrying SAM2 with LLaVA for Dense Grounded Understanding of Images and Videos
Sa2VA unifies SAM-2 segmentation with MLLM reasoning into a single model for referring segmentation and conversation on images and videos, supported by a new 72k-expression Ref-SAV dataset.
-
Uncertainty Quantification in Detection Transformers: Object-Level Calibration and Image-Level Reliability
DETRs learn an optimal specialist strategy via the Hungarian loss, motivating the new Object-level Calibration Error (OCE) metric and an image-level post-hoc uncertainty quantification framework.