2786 papers
The paper proposes GHD-DETR, a novel detection transformer architecture featuring GatedNestStage, BDB, and HAGF modules to significantly enhance object detection robustness for small, occluded, and low-contrast objects under complex imaging conditions, achieving substantial performance improvements on challenging datasets like RTTS, HazyDet, and DUO.
This paper proposes a dual-branch adaptive diffusion and motion-guided temporal alignment framework that effectively resolves the trade-off between imperceptibility and robustness in invisible video watermarking by combining content-adaptive local diffusion with coordinate-aware global anchoring to achieve superior performance against local cropping and other attacks.
This study proposes a bias-aware deep learning framework that utilizes a curated, uniformly preprocessed multisource CXR dataset and a source-prediction sanity check to mitigate dataset-origin bias, demonstrating that a customized ResNet-38 model achieves superior multi-class lung disease detection accuracy of approximately 93% compared to deeper ResNet architectures.
This paper proposes a hybrid Machine Learning-Based Security (MLBS) framework integrating K-Nearest Neighbors (KNN) and Recurrent Neural Networks (RNN) to achieve superior accuracy and robustness in detecting Distributed Denial of Service (DDoS) attacks within Internet of Things (IoT) networks, outperforming conventional models with a 98.72% detection rate.
This study proposes an adaptive multi-scale multi-expert hybrid deep learning framework that combines EfficientNetB7 and Vision Transformer with dynamic feature fusion and attention refinement to achieve state-of-the-art accuracy and explainability in breast cancer histopathological image classification on the BreakHis dataset.
This paper presents an empirical study of YouTube APIs that utilizes correlation analysis and tree-based models on 8,743 Stack Overflow interactions to identify and characterize high-risk support areas driven by environmental, code, and documentation factors, offering actionable insights for improving API support quality and response times.
This paper presents UAD-YOLO, an efficient YOLOv11-based framework enhanced with Large Separable Kernel Attention, Reparameterised Convolution, and Shape-IoU loss to achieve high-accuracy, real-time urban anomaly detection in complex scenes, validated by a new dataset and superior performance metrics over baseline models.
This study presents an explainable hierarchical machine learning framework that integrates Sequential Forward Selection with SHAP analysis to identify robust EEG biomarkers, such as frontal permutation entropy and temporal spectral features, achieving up to 81.44% accuracy in distinguishing schizophrenia patients from healthy controls.
The paper introduces NACS-Net, a novel segmentation framework that integrates a normality–deviation dual-stream, anatomy-gated iterative refinement, and a sparse concept bottleneck to effectively address foreground imbalance, low-contrast lesions, and cross-scanner generalization challenges in detecting clinically significant prostate cancer on T2-weighted MRI.
This paper proposes ODAF-YOLOv11-n, an ultra-lightweight object detector that integrates Omni-Dimensional Dynamic Convolution and a Lightweight Spatial Feature Restoration Module to achieve high-accuracy, real-time lychee detection in complex orchard environments while significantly reducing computational complexity for edge robotic deployment.