Link prediction in complex networks via fusing node centrality and local similarity indices
论文提出通过融合节点中心性和局部相似性指数来解决复杂网络中链接预测的问题，特别是在稀疏网络中的预测能力受限问题。
0 citationsRead paper
Selected work
Representative Papers
Commercial Vehicle Braking Optimization: A Robust SIFT-Trajectory Approach
To address the “zero-speed false braking” issue in commercial vehicle Automatic Emergency Braking (AEB) systems caused by CAN signal inaccuracies under low-speed conditions, this paper proposes a high-precision vehicle motion state recognition method leveraging blind-zone camera video streams. The approach introduces three key innovations: (1) a five-frame sliding-window trajectory displacement statistic; (2) a dual-threshold state decision matrix; and (3) an OBD-II–driven dynamic Region-of-Interest (ROI) configuration mechanism—collectively mitigating environmental interference and false detection of moving objects. Motion estimation employs CLAHE-enhanced image preprocessing, SIFT feature extraction, and KNN-RANSAC matching, achieving real-time processing at 14.2 ms per frame on Jetson AGX Xavier for 704×576-resolution video. Experimental results show F1-scores of 99.96% for stationary and 97.78% for moving state classification. Field deployment reduces false braking incidents by 89%, achieves 100% emergency braking success rate, and maintains system fault rate below 5%.
0 citationsRead paper
DiffFit: Disentangled Garment Warping and Texture Refinement for Virtual Try-On
Virtual try-on (VTON) faces challenges including clothing detail loss, inaccurate human-clothing alignment, inefficient inference, and poor generalization across poses and styles. This paper proposes DiffFit, a two-stage latent diffusion framework that— for the first time—decouples geometry-aware clothing deformation from texture fidelity optimization. In Stage I, a fine-grained deformation network achieves pose-adaptive geometric alignment; in Stage II, a cross-modal conditional diffusion model synthesizes high-fidelity appearance by jointly conditioning on the original garment texture, target person image, and deformed intermediate result. This progressive design significantly improves generation stability, detail preservation, and cross-pose generalization. Evaluated on large-scale benchmarks, DiffFit surpasses state-of-the-art methods in FID, LPIPS, and user studies, achieving simultaneous optimization of inference efficiency and visual quality.
0 citationsRead paper
YOLO-FDA: Integrating Hierarchical Attention and Detail Enhancement for Surface Defect Detection
Industrial surface defect detection faces challenges including diverse defect morphologies, large scale variations, strong texture interference, and difficulty in fine-grained recognition. To address these, we propose an enhanced YOLO-based multi-scale defect detection framework. Our method introduces a Detail-Directed Fusion Module (DDFM) and directional asymmetric convolution to improve sensitivity to minute defects; designs attention-weighted concatenation and cross-layer attention fusion to strengthen contextual modeling; and integrates a BiFPN architecture with hierarchical attention to optimize synergistic aggregation of low-level details and high-level semantics. Extensive experiments on multiple benchmark datasets demonstrate that our approach significantly outperforms state-of-the-art methods in mAP, small-object recall, and cross-scale robustness—achieving a favorable balance between detection accuracy and generalization capability.
0 citationsRead paper
ADA-DPM: A Neural Descriptors-based Adaptive Noise Point Filtering Strategy for SLAM
To address the fundamental trade-off between accuracy and robustness in LiDAR SLAM under dynamic object interference, point cloud noise, and unstructured environments, this paper proposes a neural-descriptor-based adaptive noise filtering SLAM framework. Our method jointly identifies dynamic points and low-quality points via a dedicated dynamic segmentation head and a global importance scoring head. To enhance geometric consistency modeling, we introduce a cross-layer graph convolutional module (GLI-GCN) that fuses multi-scale neighborhood structures. Furthermore, the framework enables end-to-end adaptive selection of high-contribution feature points and simultaneous noise suppression. Extensive experiments on multiple public benchmarks demonstrate significant improvements: average absolute trajectory error (ATE) is reduced by 23.6%, and system robustness is substantially enhanced—particularly in highly dynamic and strongly noisy scenarios. The proposed approach thus achieves superior localization accuracy without compromising reliability in challenging real-world conditions.
0 citationsRead paper
Recent publications
Latest Papers
Link prediction in complex networks via fusing node centrality and local similarity indices
论文提出通过融合节点中心性和局部相似性指数来解决复杂网络中链接预测的问题，特别是在稀疏网络中的预测能力受限问题。
0 citationsRead paper
Commercial Vehicle Braking Optimization: A Robust SIFT-Trajectory Approach
To address the “zero-speed false braking” issue in commercial vehicle Automatic Emergency Braking (AEB) systems caused by CAN signal inaccuracies under low-speed conditions, this paper proposes a high-precision vehicle motion state recognition method leveraging blind-zone camera video streams. The approach introduces three key innovations: (1) a five-frame sliding-window trajectory displacement statistic; (2) a dual-threshold state decision matrix; and (3) an OBD-II–driven dynamic Region-of-Interest (ROI) configuration mechanism—collectively mitigating environmental interference and false detection of moving objects. Motion estimation employs CLAHE-enhanced image preprocessing, SIFT feature extraction, and KNN-RANSAC matching, achieving real-time processing at 14.2 ms per frame on Jetson AGX Xavier for 704×576-resolution video. Experimental results show F1-scores of 99.96% for stationary and 97.78% for moving state classification. Field deployment reduces false braking incidents by 89%, achieves 100% emergency braking success rate, and maintains system fault rate below 5%.
0 citationsRead paper
DiffFit: Disentangled Garment Warping and Texture Refinement for Virtual Try-On
Virtual try-on (VTON) faces challenges including clothing detail loss, inaccurate human-clothing alignment, inefficient inference, and poor generalization across poses and styles. This paper proposes DiffFit, a two-stage latent diffusion framework that— for the first time—decouples geometry-aware clothing deformation from texture fidelity optimization. In Stage I, a fine-grained deformation network achieves pose-adaptive geometric alignment; in Stage II, a cross-modal conditional diffusion model synthesizes high-fidelity appearance by jointly conditioning on the original garment texture, target person image, and deformed intermediate result. This progressive design significantly improves generation stability, detail preservation, and cross-pose generalization. Evaluated on large-scale benchmarks, DiffFit surpasses state-of-the-art methods in FID, LPIPS, and user studies, achieving simultaneous optimization of inference efficiency and visual quality.
0 citationsRead paper
YOLO-FDA: Integrating Hierarchical Attention and Detail Enhancement for Surface Defect Detection
Industrial surface defect detection faces challenges including diverse defect morphologies, large scale variations, strong texture interference, and difficulty in fine-grained recognition. To address these, we propose an enhanced YOLO-based multi-scale defect detection framework. Our method introduces a Detail-Directed Fusion Module (DDFM) and directional asymmetric convolution to improve sensitivity to minute defects; designs attention-weighted concatenation and cross-layer attention fusion to strengthen contextual modeling; and integrates a BiFPN architecture with hierarchical attention to optimize synergistic aggregation of low-level details and high-level semantics. Extensive experiments on multiple benchmark datasets demonstrate that our approach significantly outperforms state-of-the-art methods in mAP, small-object recall, and cross-scale robustness—achieving a favorable balance between detection accuracy and generalization capability.
0 citationsRead paper
ADA-DPM: A Neural Descriptors-based Adaptive Noise Point Filtering Strategy for SLAM
To address the fundamental trade-off between accuracy and robustness in LiDAR SLAM under dynamic object interference, point cloud noise, and unstructured environments, this paper proposes a neural-descriptor-based adaptive noise filtering SLAM framework. Our method jointly identifies dynamic points and low-quality points via a dedicated dynamic segmentation head and a global importance scoring head. To enhance geometric consistency modeling, we introduce a cross-layer graph convolutional module (GLI-GCN) that fuses multi-scale neighborhood structures. Furthermore, the framework enables end-to-end adaptive selection of high-contribution feature points and simultaneous noise suppression. Extensive experiments on multiple public benchmarks demonstrate significant improvements: average absolute trajectory error (ATE) is reduced by 23.6%, and system robustness is substantially enhanced—particularly in highly dynamic and strongly noisy scenarios. The proposed approach thus achieves superior localization accuracy without compromising reliability in challenging real-world conditions.
0 citationsRead paper