2833 papers
This paper introduces LarvaFormer, a novel hybrid deep learning framework combining CNNs and Vision Transformers with specialized attention mechanisms, which achieves state-of-the-art accuracy in automated mosquito larval classification to enable rapid and environmentally sustainable disease control.
This paper proposes a lightweight, high-accuracy surface defect detection model for metal strips, named YOLOv5s-SC2, which integrates architectural improvements like SAC and RepGFPN to achieve 84.3% mAP on the NEU-DET dataset and utilizes LAMP pruning to significantly reduce model size and computational cost for efficient edge deployment with minimal accuracy loss.
PolicyFaultBench is a mutation-based benchmark that validates tool-using AI agents by rigorously testing runtime policy mediation and proposal-interface conformance, revealing that while agents can achieve high execution success, they may still fail strict acceptance criteria due to subtle interface deviations that require targeted probes to detect.
This paper introduces YOLO11m-opt, an optimized object detection model that achieves high-precision identification of 36 parasitoid wasp species from low-cost microscopy images by incorporating specialized architectural enhancements and a custom dataset, thereby enabling automated, expert-independent taxonomic analysis for agricultural and ecological applications.
This paper proposes HRL-TSCH-ERC, a hierarchical reinforcement learning-based scheduling method that dynamically allocates resources and schedules links in karst cave Wireless Mesh Networks to significantly improve packet delivery, reliability, and energy efficiency for heterogeneous emergency rescue services compared to existing algorithms.
HierGait is a hierarchical multi-modal gait recognition framework that integrates adaptive multi-scale temporal enhancement, structure-guided alignment fusion, and global context recalibration to overcome limitations in fine-grained motion capture, cross-modal alignment, and temporal modeling, achieving state-of-the-art performance on the CASIA-B dataset.
EDD-3DGS is an illumination-guided depth distillation framework that overcomes low-light reconstruction failures in 3D Gaussian Splatting by dynamically balancing photometric and depth supervision, filtering noisy priors, and aligning monocular depth scales to eliminate floating artifacts while preserving geometric fidelity.
This paper presents a prior-constrained lightweight HRNet framework optimized via Quantum-behaved Particle Swarm Optimization that achieves high-accuracy, efficient IHDI classification for Developmental Dysplasia of the Hip by integrating geometric priors, model compression, and adaptive hyperparameter tuning for clinical decision support.
This study proposes a corpus-scale constrained optimization framework that leverages latent slip properties and excavation context to significantly improve the automated reassembly of fragmented bamboo slips, outperforming traditional pairwise matching methods in both accuracy and error reduction.
This paper introduces the Quantum General Public License (QGPL) v3.0, which mandates post-quantum cryptographic integrity for open science distributions, and demonstrates its practical feasibility through a hybrid signature scheme that ensures long-term artifact authenticity on commodity hardware without centralized infrastructure.