Guangping Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Guangping Li has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Bingo Wing-Kuen Ling
Faculty of Information Engineering, Guangdong University of Technology, Guangzhou, China
6 shared publications - Dingkai Liang
School of Information Engineering, Guangdong University of Technology, Guangzhou, China
3 shared publications - Huanling Xiao
School of Information Engineering, Guangdong University of Technology, Guangzhou, China
3 shared publications Bingo Wing-Kuen Ling
Guangdong University of Technology, Guangzhou, China
2 shared publications- Chenghui Liu
School of Information Engineering, Guangdong University of Technology, Guangzhou, China
2 shared publications - Dingkai Liang
School of Information Engineering, Guangdong University of Technology, Guangzhou, China
2 shared publications
Latest Publications
- 24 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 26  (7) , 10768-10777
MTCLnet: A Computationally Efficient and Adaptive Information Fusion Network for Cross-Modal Point Cloud Completion
Attention-based multi-modal fusion frameworks have significantly advanced View-guided Point Cloud Completion (ViPC), which aims to recover a complete 3D point cloud from a partial one with the aid of a single-view image. However, existing methods commonly employ asymmetric cross-attention for point cloud and image feature fusion, in which one modality is designated as the query and the other as key-value pairs, resulting in an information flow biased toward the querying modality rather than enabling flexible feature extraction across modalities. Moreover, supplement completion loss that require the participation of all point cloud and image tokens introduce substantial computational overhead. To address these limitations, we propose Memory Transfer with Contrastive Learning Network termed MTCLnet, a novel framework that leverages contrastive learning to effectively align the feature spaces between point cloud and image modalities. Furthermore, we design a cross-attention-based module equipped with learnable tokens to adaptively fuse informative features across modalities. The integrated design enables more accurate and computationally efficient point cloud completion. Extensive experiments on the ShapeNet-ViPC dataset demonstrate that our method outperforms existing approaches, achieving both improved reconstruction quality and reduced computational complexity.
- 24 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 26  (7) , 10768-10777
MTCLnet: A Computationally Efficient and Adaptive Information Fusion Network for Cross-Modal Point Cloud Completion
Attention-based multi-modal fusion frameworks have significantly advanced View-guided Point Cloud Completion (ViPC), which aims to recover a complete 3D point cloud from a partial one with the aid of a single-view image. However, existing methods commonly employ asymmetric cross-attention for point cloud and image feature fusion, in which one modality is designated as the query and the other as key-value pairs, resulting in an information flow biased toward the querying modality rather than enabling flexible feature extraction across modalities. Moreover, supplement completion loss that require the participation of all point cloud and image tokens introduce substantial computational overhead. To address these limitations, we propose Memory Transfer with Contrastive Learning Network termed MTCLnet, a novel framework that leverages contrastive learning to effectively align the feature spaces between point cloud and image modalities. Furthermore, we design a cross-attention-based module equipped with learnable tokens to adaptively fuse informative features across modalities. The integrated design enables more accurate and computationally efficient point cloud completion. Extensive experiments on the ShapeNet-ViPC dataset demonstrate that our method outperforms existing approaches, achieving both improved reconstruction quality and reduced computational complexity.
- Open Access
- 11 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 26647-26661
Multi-Class Diagnosis of Pulmonary Hypertension in the Presence of COPD via CT Imaging and Transformer–CNN Fusion
Pulmonary hypertension (PH) is a life-threatening condition that often coexists with chronic obstructive pulmonary disease (COPD), yet few diagnostic studies explicitly account for this comorbidity. When diagnosing PH from computed tomography (CT) images using deep learning models, irrelevant visual components commonly present in the imaging field may introduce background noise and hinder accurate assessment of the pulmonary artery. To tackle these issues, we construct a CT-based dataset comprising four categories: PH with COPD, PH only, COPD only, and controls.We propose a novel classification framework that integrates EfficientNet-B0 and a modified Swin Transformer branch enhanced with Multiscale Dynamic Swin Transformer (MDST) blocks. To guide the model’s attention toward diagnostically relevant regions, we introduce a Pulmonary Artery Region Masking (PARM) strategy that adaptively isolates the pulmonary artery and suppresses background noise. Experimental results show that the PARM-guided EfficientNet combined with the Multiscale Dynamic Swin Transformer (Eff-MDST) achieves an accuracy of 0.8569, a macro-precision of 0.8634, a macro-recall of 0.8601, and a macro-F1 score of 0.8570. These results clearly outperform conventional Convolutional Neural Networks (CNNs) as well as baseline Transformer architectures, with an overall accuracy improvement of at least 0.0147. These findings highlight the potential of combining anatomical priors with Transformer-based architectures to advance CT-based PH diagnosis in the presence of COPD.
- 11 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 26647-26661
Open AccessMulti-Class Diagnosis of Pulmonary Hypertension in the Presence of COPD via CT Imaging and Transformer–CNN Fusion
Pulmonary hypertension (PH) is a life-threatening condition that often coexists with chronic obstructive pulmonary disease (COPD), yet few diagnostic studies explicitly account for this comorbidity. When diagnosing PH from computed tomography (CT) images using deep learning models, irrelevant visual components commonly present in the imaging field may introduce background noise and hinder accurate assessment of the pulmonary artery. To tackle these issues, we construct a CT-based dataset comprising four categories: PH with COPD, PH only, COPD only, and controls.We propose a novel classification framework that integrates EfficientNet-B0 and a modified Swin Transformer branch enhanced with Multiscale Dynamic Swin Transformer (MDST) blocks. To guide the model’s attention toward diagnostically relevant regions, we introduce a Pulmonary Artery Region Masking (PARM) strategy that adaptively isolates the pulmonary artery and suppresses background noise. Experimental results show that the PARM-guided EfficientNet combined with the Multiscale Dynamic Swin Transformer (Eff-MDST) achieves an accuracy of 0.8569, a macro-precision of 0.8634, a macro-recall of 0.8601, and a macro-F1 score of 0.8570. These results clearly outperform conventional Convolutional Neural Networks (CNNs) as well as baseline Transformer architectures, with an overall accuracy improvement of at least 0.0147. These findings highlight the potential of combining anatomical priors with Transformer-based architectures to advance CT-based PH diagnosis in the presence of COPD.
- Open Access
MogaDepth: Multi-Order Feature Hierarchy Fusion for Lightweight Monocular Depth Estimation
Monocular depth estimation is a fundamental task with broad applications in autonomous driving and augmented reality. While recent lightweight methods achieve impressive performance, they often neglect the interaction of mid-order semantic features, which are crucial for capturing object structures and spatial relationships that directly impact depth accuracy. To address this limitation, we propose MogaDepth, a lightweight yet expressive architecture. It introduces a novel Continuous Multi-Order Gated Aggregation (CMOGA) module that explicitly enhances mid-level feature representations through multi-order receptive fields. In addition, we present MambaSync, a global–local interaction unit that enables efficient feature communication across different contexts. Extensive experiments demonstrate that MogaDepth achieves highly competitive or superior performance on KITTI, improving key error metrics while maintaining comparable model size. On the Make3D benchmark, it consistently outperforms existing methods, showing strong robustness to domain shifts and challenging scenarios such as low-texture regions. Moreover, MogaDepth achieves an improved trade-off between accuracy and efficiency, running up to 13% faster on edge devices without compromising performance. These results establish MogaDepth as an effective and efficient solution for real-world monocular depth estimation.
Open AccessMogaDepth: Multi-Order Feature Hierarchy Fusion for Lightweight Monocular Depth Estimation
Monocular depth estimation is a fundamental task with broad applications in autonomous driving and augmented reality. While recent lightweight methods achieve impressive performance, they often neglect the interaction of mid-order semantic features, which are crucial for capturing object structures and spatial relationships that directly impact depth accuracy. To address this limitation, we propose MogaDepth, a lightweight yet expressive architecture. It introduces a novel Continuous Multi-Order Gated Aggregation (CMOGA) module that explicitly enhances mid-level feature representations through multi-order receptive fields. In addition, we present MambaSync, a global–local interaction unit that enables efficient feature communication across different contexts. Extensive experiments demonstrate that MogaDepth achieves highly competitive or superior performance on KITTI, improving key error metrics while maintaining comparable model size. On the Make3D benchmark, it consistently outperforms existing methods, showing strong robustness to domain shifts and challenging scenarios such as low-texture regions. Moreover, MogaDepth achieves an improved trade-off between accuracy and efficiency, running up to 13% faster on edge devices without compromising performance. These results establish MogaDepth as an effective and efficient solution for real-world monocular depth estimation.
- Open Access
- 13 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19  (19391404) , 5021-5034
AMFC-DEIM: Improved DEIM With Adaptive Matching and Focal Convolution for Remote Sensing Small Object Detection
While Convolutional Neural Network (CNN)-based methods for small object detection in remote sensing imagery have advanced considerably, substantial challenges remain unresolved, primarily stemming from complex backgrounds and insufficient feature representation. To address these issues, we propose a novel architecture specifically designed to accommodate the unique demands of small objects, termed AMFC-DEIM. This framework introduces three key innovations: (1) the Adaptive One-to-One (O2O) matching mechanism, which enhances Dense O2O matching by adaptively adjusting the matching grid configuration to the object distribution, thereby preserving the resolution of small objects throughout training; (2) the Focal Convolution Module, engineered to explicitly align with the spatial characteristics of small objects for extracting fine-grained features; and (3) the Enhanced Normalized Wasserstein Distance, which stabilizes the training process and bolsters performance on small targets. Comprehensive experiments conducted on three benchmark remote sensing small object detection datasets: RSOD, LEVIR-SHIP and NWPU VHR-10, demonstrate that AMFC-DEIM achieves remarkable performance, attaining AP$_{50}$ scores of 96.2%, 86.2%, and 95.1%, respectively, while maintaining only 5.27M parameters. These results substantially outperform several established benchmark models and state-of-the-art methods.
- 13 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19  (19391404) , 5021-5034
Open AccessAMFC-DEIM: Improved DEIM With Adaptive Matching and Focal Convolution for Remote Sensing Small Object Detection
While Convolutional Neural Network (CNN)-based methods for small object detection in remote sensing imagery have advanced considerably, substantial challenges remain unresolved, primarily stemming from complex backgrounds and insufficient feature representation. To address these issues, we propose a novel architecture specifically designed to accommodate the unique demands of small objects, termed AMFC-DEIM. This framework introduces three key innovations: (1) the Adaptive One-to-One (O2O) matching mechanism, which enhances Dense O2O matching by adaptively adjusting the matching grid configuration to the object distribution, thereby preserving the resolution of small objects throughout training; (2) the Focal Convolution Module, engineered to explicitly align with the spatial characteristics of small objects for extracting fine-grained features; and (3) the Enhanced Normalized Wasserstein Distance, which stabilizes the training process and bolsters performance on small targets. Comprehensive experiments conducted on three benchmark remote sensing small object detection datasets: RSOD, LEVIR-SHIP and NWPU VHR-10, demonstrate that AMFC-DEIM achieves remarkable performance, attaining AP$_{50}$ scores of 96.2%, 86.2%, and 95.1%, respectively, while maintaining only 5.27M parameters. These results substantially outperform several established benchmark models and state-of-the-art methods.
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19, 36308-36326
DviCD: Lightweight Multiscale Calibration for High-Resolution Remote Sensing Change Detection
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19, 36308-36326
Open AccessDviCD: Lightweight Multiscale Calibration for High-Resolution Remote Sensing Change Detection
- Open Access
- 9 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18  (19391404) , 24144-24161
UGD-DLinkNet: An Enhanced Network for Occluded Road Extraction Using Attention Mechanisms and Uncertainty Estimation
Road information extracted from high-resolution remote sensing images is crucial for urban planning and traffic navigation. However, challenges such as complex and variable geometric and contextual features of roads, occlusion by trees and buildings, and labeling errors in datasets present significant difficulties. To address these issues, we propose the UGD-DLinkNet model. Specifically, attention mechanisms are employed to enhance road feature extraction in occluded and complex regions by adaptively focusing on informative spatial and channel-wise cues, while model uncertainty estimation helps reduce the impact of noisy annotations by guiding learning toward more reliable predictions. First, we introduce a hybrid attention module to strengthen the encoder's ability to capture essential features and structural information. Second, a self-attention unit is integrated into the bridging network to form a dilated convolution attention module (DCAM), improving the perception of semantic features across scales. Third, a channel attention module refines skip connections, bridging shallow and deep semantic information. Finally, we incorporate Monte Carlo dropout (MC dropout) and propose an uncertainty-guided knowledge distillation strategy to mitigate labeling noise. Experiments conducted on the Massachusetts, CHN6-CUG, and DeepGlobe benchmark datasets demonstrate the comprehensive superiority of UGD-DLinkNet. On the Massachusetts dataset, our model achieves the highest F1-score of $78.75\%$. On CHN6-CUG, it obtains the best overall performance with an F1-score of $76.50\%$ and an IoU of $61.95\%$, exceeding the best results among recent mainstream methods by $4.68\%$ and $2.33\%$, respectively. On DeepGlobe, UGD-DLinkNet also achieves top performance, with an F1-score of $81.49\%$ and an IoU of $68.77\%$, highlighting its strong generalization and robustness across diverse scenarios. The code will be available at https://github.com/VisionENet/UGD-DLinkNet.
- 9 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18  (19391404) , 24144-24161
Open AccessUGD-DLinkNet: An Enhanced Network for Occluded Road Extraction Using Attention Mechanisms and Uncertainty Estimation
Road information extracted from high-resolution remote sensing images is crucial for urban planning and traffic navigation. However, challenges such as complex and variable geometric and contextual features of roads, occlusion by trees and buildings, and labeling errors in datasets present significant difficulties. To address these issues, we propose the UGD-DLinkNet model. Specifically, attention mechanisms are employed to enhance road feature extraction in occluded and complex regions by adaptively focusing on informative spatial and channel-wise cues, while model uncertainty estimation helps reduce the impact of noisy annotations by guiding learning toward more reliable predictions. First, we introduce a hybrid attention module to strengthen the encoder's ability to capture essential features and structural information. Second, a self-attention unit is integrated into the bridging network to form a dilated convolution attention module (DCAM), improving the perception of semantic features across scales. Third, a channel attention module refines skip connections, bridging shallow and deep semantic information. Finally, we incorporate Monte Carlo dropout (MC dropout) and propose an uncertainty-guided knowledge distillation strategy to mitigate labeling noise. Experiments conducted on the Massachusetts, CHN6-CUG, and DeepGlobe benchmark datasets demonstrate the comprehensive superiority of UGD-DLinkNet. On the Massachusetts dataset, our model achieves the highest F1-score of $78.75\%$. On CHN6-CUG, it obtains the best overall performance with an F1-score of $76.50\%$ and an IoU of $61.95\%$, exceeding the best results among recent mainstream methods by $4.68\%$ and $2.33\%$, respectively. On DeepGlobe, UGD-DLinkNet also achieves top performance, with an F1-score of $81.49\%$ and an IoU of $68.77\%$, highlighting its strong generalization and robustness across diverse scenarios. The code will be available at https://github.com/VisionENet/UGD-DLinkNet.
- Open Access
- 16 February 2025
- Published by MDPI AG in Electronics
- Vol. 14  (4) , 767
A Hard Negatives Mining and Enhancing Method for Multi-Modal Contrastive Learning
Contrastive learning has emerged as a dominant paradigm for understanding 3D open-world environments, particularly in the realm of multi-modalities. However, due to the nature of self-supervised learning and the limited size of 3D datasets, pre-trained models in the 3D point cloud domain often suffer from overfitting in downstream tasks, especially in zero-shot classification. To tackle this problem, we design a module to mine and enhance hard negatives from datasets, which are useful to improve the discrimination of models. This module could be seamlessly integrated into cross-modal contrastive learning frameworks, addressing the overfitting issue by enhancing the mined hard negatives during the process of training. This module consists of two key components: mining and enhancing. In the process of mining, we identify hard negative samples by examining similarity relationships between vision–vision and vision–text modalities, locating hard negative pairs within the visual domain. In the process of enhancing, we compute weighting coefficients via the similarity differences of these mined hard negatives. By enhancing the mined hard negatives while leaving others unchanged, we improve the overall performance and discrimination of models. A series of experiments demonstrate that our module can be easily incorporated into various contrastive learning frameworks, leading to improved model performance in both zero-shot and few-shot tasks.
- 16 February 2025
- Published by MDPI AG in Electronics
- Vol. 14  (4) , 767
Open AccessA Hard Negatives Mining and Enhancing Method for Multi-Modal Contrastive Learning
Contrastive learning has emerged as a dominant paradigm for understanding 3D open-world environments, particularly in the realm of multi-modalities. However, due to the nature of self-supervised learning and the limited size of 3D datasets, pre-trained models in the 3D point cloud domain often suffer from overfitting in downstream tasks, especially in zero-shot classification. To tackle this problem, we design a module to mine and enhance hard negatives from datasets, which are useful to improve the discrimination of models. This module could be seamlessly integrated into cross-modal contrastive learning frameworks, addressing the overfitting issue by enhancing the mined hard negatives during the process of training. This module consists of two key components: mining and enhancing. In the process of mining, we identify hard negative samples by examining similarity relationships between vision–vision and vision–text modalities, locating hard negative pairs within the visual domain. In the process of enhancing, we compute weighting coefficients via the similarity differences of these mined hard negatives. By enhancing the mined hard negatives while leaving others unchanged, we improve the overall performance and discrimination of models. A series of experiments demonstrate that our module can be easily incorporated into various contrastive learning frameworks, leading to improved model performance in both zero-shot and few-shot tasks.
- 1 January 2025
- Published by SPIE-Intl Soc Optical Eng in Journal of Applied Remote Sensing
- Vol. 19  (01) , 016515-21
MGLFA-Net: multi-scale global–local feature aggregation network for remote sensing change detection
Change detection (CD) is a critical research area in the field of remote sensing (RS), involving the identification of altered regions by comparing multi-temporal RS images of the same location collected at different times. CD is widely applied in real-world scenarios, including landslide detection,1 building change detection,2 forest change monitoring,3 and urban growth analysis.4 Detecting land cover and structural changes using RS images is, therefore, highly significant for monitoring changes on the Earth’s surface.5
- 1 January 2025
- Published by SPIE-Intl Soc Optical Eng in Journal of Applied Remote Sensing
- Vol. 19  (01) , 016515-21
MGLFA-Net: multi-scale global–local feature aggregation network for remote sensing change detection
Change detection (CD) is a critical research area in the field of remote sensing (RS), involving the identification of altered regions by comparing multi-temporal RS images of the same location collected at different times. CD is widely applied in real-world scenarios, including landslide detection,1 building change detection,2 forest change monitoring,3 and urban growth analysis.4 Detecting land cover and structural changes using RS images is, therefore, highly significant for monitoring changes on the Earth’s surface.5
- 4 March 2024
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83  (33) , 79093-79115
AFpoint: adaptively fusing local and global features for point cloud
- 4 March 2024
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83  (33) , 79093-79115
AFpoint: adaptively fusing local and global features for point cloud
- 1 March 2024
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83  (33) , 79805-79814
Multi-scale cross-fusion for arbitrary scale image super resolution
- 1 March 2024
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83  (33) , 79805-79814
Multi-scale cross-fusion for arbitrary scale image super resolution