DEtection TRansformer
PulseAugur coverage of DEtection TRansformer — every cluster mentioning DEtection TRansformer across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
LiG-DETR framework enhances aerial object detection with local-global feature integration
Researchers have introduced LiG-DETR, a novel framework designed to improve aerial object detection by addressing the challenges of scale and density variations. The method reformulates image slicing into high-fidelity …
-
Detection Transformers applied to Diffusion MRI for microstructure quantification
Researchers have developed a novel approach to quantify white matter microstructure in diffusion MRI by reframing the problem as an object detection task. This method utilizes the Detection Transformer (DETR) architectu…
-
SeGDeP enhances reasoning segmentation by decoupling semantic and geometric prompts
Researchers have developed SeGDeP, a novel interface for reasoning segmentation that disentangles semantic understanding from spatial localization. This approach uses separate branches for semantic prompts and geometric…
-
New framework exploits stability-plasticity in pretrained detectors for incremental object detection
Researchers have developed a new framework for incremental object detection that leverages the stability-plasticity asymmetry found in pretrained DETR-based detectors. This approach freezes localization heads to maintai…
-
New ChessQueries method achieves 99.2% accuracy in chess board recognition
Researchers have developed ChessQueries, a novel method for recognizing chess board states from images. This new approach, which combines a ViT encoder with a DETR-style decoder, significantly improves upon existing ben…
-
New research advances multi-object tracking with 3D geometry and LLM integration · 4 sources tracked
Researchers have developed new methods for multi-object tracking in videos, aiming to improve accuracy and efficiency. PLANET, a new end-to-end tracker, moves beyond image-plane limitations by incorporating 3D scene geo…
-
New CLSC DETR model enhances small object detection for UAVs
Researchers have developed a new object detection model called CLSC DETR, designed to improve the identification of small objects in complex aerial scenes captured by unmanned aerial vehicles (UAVs). This model addresse…
-
New framework boosts industrial defect detection with continuity learning
Researchers have developed a new continuity-driven representation learning framework to improve industrial defect detection. This method leverages normal-dominant regions as dense auxiliary supervision, introducing two …
-
AI models compared for military target detection in new arXiv study
A new research paper published on arXiv details a comparative study of various out-of-the-box technologies for automatic target detection and recognition (ATD/R). The study benchmarks several YOLO iterations and DETR fr…
-
New research explores set decoder performance in computer vision
Researchers have developed a new method for analyzing set decoders in computer vision, focusing on the tension between improving individual predictions and maintaining the overall utility of the prediction set. Their st…
-
MammoMix uses Mixture-of-Experts for robust mammogram breast detection
Researchers have developed MammoMix, a new framework utilizing the Mixture-of-Experts (MoE) paradigm to improve the detection of breast lesions in mammograms. This approach trains individual expert models on specific da…
-
New Poly-DETR model bridges object detection and segmentation
Researchers have introduced Polygon Detection Transformers (Poly-DETR), a novel approach that bridges the gap between object detection and segmentation. This method utilizes a polar representation to directly construct …
-
Modern backbones boost AI for mammography classification and lesion localization
Researchers have explored the use of modern neural network backbones within a multi-task DETR framework to enhance mammography classification and lesion localization. The study found that advanced backbones like ConvNeX…
-
New SkySeaLand benchmark targets satellite object detection challenges
Researchers have introduced SkySeaLand, a new benchmark dataset designed for satellite object detection, particularly focusing on wide-format scenes and small targets. The dataset comprises 1,307 high-resolution satelli…
-
Deep learning model advances utility pole and sign detection
Researchers have developed a deep learning framework for detecting, segmenting, and estimating the lean angle of utility poles, as well as classifying attached warning signs. The system, based on a modified Detection Tr…
-
New open dataset OSSDD released for Sentinel-1 ship detection
Researchers have introduced OSSDD, a new open dataset designed for training neural networks to detect ships in Synthetic Aperture Radar (SAR) images. This dataset, built upon the OpenSARShip 1.0 dataset, provides 15,197…
-
2D Detection Transformers Show Surprising 3D Understanding
Researchers have investigated the 3D object-level understanding capabilities of pre-trained 2D detection transformers, such as DETR. Their findings indicate that these models, despite being trained solely on 2D data wit…
-
RiO-DETR advances real-time oriented object detection with novel transformer architecture
Researchers have introduced RiO-DETR, a novel detection transformer designed for real-time oriented object detection. This model addresses key challenges in adapting existing transformer architectures to oriented boundi…
-
Gaze-DETR uses priority maps for infrared UAV detection
Researchers have developed Gaze-DETR, a novel approach for detecting small, weak targets in infrared imagery, particularly for Unmanned Aerial Vehicles (UAVs). This method incorporates a bio-inspired internal priority m…
-
AI-powered virtual white cane uses LiDAR and DETR
A research paper details the development of VirtualCane, an AI-powered virtual white cane designed to assist visually impaired individuals. The system utilizes technologies like LiDAR and DETR, achieving performance rat…