Haoyu Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Haoyu Wang has a total of 29 co-authors. Here are the top co-authors by number of shared publications:
- Lei Zhang
Northwestern Polytechnical University, Xi'an, China
3 shared publications - Wei Wei
Northwestern Polytechnical University, Xi'an, China
3 shared publications - Yanning Zhang
Northwestern Polytechnical University
2 shared publications - Airong Qian
Laboratory for Bone Metabolism, Xi’an Key Laboratory of Special Medicine and Health Engineering, Key Laboratory for Space Biosciences and Biotechnology, Research Center for Special Medicine and Health Systems Engineering, NPU-UAB Joint Laboratory for Bone Metabolism, School of Life Sciences, Northwestern Polytechnical University, Xi’an 710072, China
1 shared publication - Chen Ding
School of Computer Science and Technology, Xi’an University of Posts and Telecommunications, Xi’an 710121, China
1 shared publication - Chu Chen
Department of Cellular and Integrative Physiology, Joe R. & Teresa Lozano Long School of Medicine, University of Texas at San Antonio Health Science Center, San Antonio, Texas, USA
1 shared publication
Latest Publications
- Open Access
- 24 June 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) , 17671-17686
Text-Guided Distribution Calibration for Few-Shot Object Detection in Remote Sensing Images
In recent years, few-shot object detection(FSOD) in remote sensing images(RSIs) has received increasing attention. However, due to the large difference in the number of labeled samples between the base classes and the novel classes, using only visual information for object detection will cause the features learned by the model to be biased towards the base classes, resulting in poor generalization ability in the novel classes with scarce labeled samples. In this paper, we propose a Text-guided distribution calibration network for few-shot object detection on RSIs. Considering the limited visual information of the novel classes, we propose a cross-modal knowledge transfer strategy, which aims to extract the corresponding text feature of the object class name through a multi-modal pre-training model CLIP and transfer the text knowledge to the FSOD model, to mitigate the feature bias problem. Following this idea, we design a text-guided distribution calibration module(TDCM), for each query image which utilizes the intra-image object class distribution defined by the text features to calibrate the object class distribution computed based on the visual features using a knowledge distillation loss for model training. By doing this, the cross-class transferable text knowledge can be transferred to regularize the learned visual features step siding bias on base classes and thus improve the generalization capacity. We conducted experiments on the NWPU VHR-10 and DIOR datasets and clarified the superior performance of the proposed method compared with several state-of-the-art comparison methods.
- 24 June 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) , 17671-17686
Open AccessText-Guided Distribution Calibration for Few-Shot Object Detection in Remote Sensing Images
In recent years, few-shot object detection(FSOD) in remote sensing images(RSIs) has received increasing attention. However, due to the large difference in the number of labeled samples between the base classes and the novel classes, using only visual information for object detection will cause the features learned by the model to be biased towards the base classes, resulting in poor generalization ability in the novel classes with scarce labeled samples. In this paper, we propose a Text-guided distribution calibration network for few-shot object detection on RSIs. Considering the limited visual information of the novel classes, we propose a cross-modal knowledge transfer strategy, which aims to extract the corresponding text feature of the object class name through a multi-modal pre-training model CLIP and transfer the text knowledge to the FSOD model, to mitigate the feature bias problem. Following this idea, we design a text-guided distribution calibration module(TDCM), for each query image which utilizes the intra-image object class distribution defined by the text features to calibrate the object class distribution computed based on the visual features using a knowledge distillation loss for model training. By doing this, the cross-class transferable text knowledge can be transferred to regularize the learned visual features step siding bias on base classes and thus improve the generalization capacity. We conducted experiments on the NWPU VHR-10 and DIOR datasets and clarified the superior performance of the proposed method compared with several state-of-the-art comparison methods.
- 30 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Big Data
- Vol. 11 (3) , 1013-1026
Adapt Anything: Tailor Any Image Classifier Across Domains and Categories Using Text-to-Image Diffusion Models
We study a novel problem in this paper, that is, if a modern text-to-image diffusion model can tailor any image classifier across domains and categories. Existing domain adaption works exploit both source and target data for domain alignment so as to transfer the knowledge from the labeled source data to the unlabeled target data. However, as the development of text-to-image diffusion models, we wonder if the high-fidelity synthetic data can serve as a surrogate of the source data in real world. In this way, we do not need to collect and annotate the source data for each image classification task in a one-for-one manner. Instead, we utilize only one off-the-shelf text-to-image model to synthesize images with labels derived from text prompts, and then leverage them as a bridge to dig out the knowledge from the task-agnostic text-to-image generator to the task-oriented image classifier via domain adaptation. Such a one-for-all adaptation paradigm allows us to adapt anything in the world using only one text-to-image generator as well as any unlabeled target data. Extensive experiments validate the feasibility of this idea, which even surprisingly surpasses the state-of-the-art domain adaptation works using the source data collected and annotated in real world.
- 30 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Big Data
- Vol. 11 (3) , 1013-1026
Adapt Anything: Tailor Any Image Classifier Across Domains and Categories Using Text-to-Image Diffusion Models
We study a novel problem in this paper, that is, if a modern text-to-image diffusion model can tailor any image classifier across domains and categories. Existing domain adaption works exploit both source and target data for domain alignment so as to transfer the knowledge from the labeled source data to the unlabeled target data. However, as the development of text-to-image diffusion models, we wonder if the high-fidelity synthetic data can serve as a surrogate of the source data in real world. In this way, we do not need to collect and annotate the source data for each image classification task in a one-for-one manner. Instead, we utilize only one off-the-shelf text-to-image model to synthesize images with labels derived from text prompts, and then leverage them as a bridge to dig out the knowledge from the task-agnostic text-to-image generator to the task-oriented image classifier via domain adaptation. Such a one-for-all adaptation paradigm allows us to adapt anything in the world using only one text-to-image generator as well as any unlabeled target data. Extensive experiments validate the feasibility of this idea, which even surprisingly surpasses the state-of-the-art domain adaptation works using the source data collected and annotated in real world.
- Open Access
- 1 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 7507-7516
Glocal Energy-based Learning for Few-Shot Open-Set Recognition
Few-shot open-set recognition (FSOR) is a challenging task of great practical value. It aims to categorize a sample to one of the predefined, closed-set classes illustrated by few examples while being able to reject the sample from unknown classes. In this work, we approach the FSOR task by proposing a novel energy-based hybrid model. The model is composed of two branches, where a classification branch learns a metric to classify a sample to one of closed-set classes and the energy branch explicitly estimates the open-set probability. To achieve holistic detection of open-set samples, our model leverages both class-wise and pixel-wise features to learn a glocal energy-based score, in which a global energy score is learned using the class-wise features, while a local energy score is learned using the pixel-wise features. The model is enforced to assign large energy scores to samples that are deviated from the few-shot examples in either the class-wise features or the pixel-wise features, and to assign small energy scores otherwise. Experiments on three standard FSOR datasets show the superior performance of our model. 1 1 Code is available at https://github.com/00why00/Glocal
- 1 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 7507-7516
Open AccessGlocal Energy-based Learning for Few-Shot Open-Set Recognition
Few-shot open-set recognition (FSOR) is a challenging task of great practical value. It aims to categorize a sample to one of the predefined, closed-set classes illustrated by few examples while being able to reject the sample from unknown classes. In this work, we approach the FSOR task by proposing a novel energy-based hybrid model. The model is composed of two branches, where a classification branch learns a metric to classify a sample to one of closed-set classes and the energy branch explicitly estimates the open-set probability. To achieve holistic detection of open-set samples, our model leverages both class-wise and pixel-wise features to learn a glocal energy-based score, in which a global energy score is learned using the class-wise features, while a local energy score is learned using the pixel-wise features. The model is enforced to assign large energy scores to samples that are deviated from the few-shot examples in either the class-wise features or the pixel-wise features, and to assign small energy scores otherwise. Experiments on three standard FSOR datasets show the superior performance of our model. 1 1 Code is available at https://github.com/00why00/Glocal
- Open Access
- 4 January 2018
- Published by Spandidos Publications in Experimental and Therapeutic Medicine
- Vol. 15 (3) , 2269-2276
Limethason reduces airway inflammation in a murine model of ovalbumin-induced chronic asthma without causing side effects
Airway inflammation is the major pathological feature of asthma. Thus, the current therapeutic strategy for asthma is to control inflammation. Limethason, an anti‑inflammation drug, is widely used in rheumatoid arthritis treatment. The aim of the present study was to detect the anti‑inflammatory effect and side effects of limethason on airways that were sensitized with ovalbumin in a murine model of chronic asthma. In the present study, BALB/c mice were sensitized with ovalbumin. Airway hyperresponsiveness was estimated, and hematoxylin and eosin staining, Periodic acid‑Schiff staining and bronchoalveolar lavage were used to detect the effect on chronic asthma. Limethason effectively reduced airway hyperresponsiveness, and inhibited inflammatory cell infiltration and mucus secretion. Bronchoalveolar lavage fluid analysis revealed that limethason suppressed levels of airway eosinophils. In the period of treatment, limethason exhibited no influence on morphology of the femoral head, bone mineral content or bone mineral density, which were detected by histological studies and dual‑energy X‑ray absorptiometry. The index of liver, spleen, kidney, gastrocnemius and brown adipose tissue also demonstrated that limethason had no adverse effects on organs and tissues. The present study revealed that limethason could effectively reduce inflammation in an asthma mouse model without side effects. Therefore, limethason may have therapeutic potential for treating chronic asthma clinically.
- 4 January 2018
- Published by Spandidos Publications in Experimental and Therapeutic Medicine
- Vol. 15 (3) , 2269-2276
Open AccessLimethason reduces airway inflammation in a murine model of ovalbumin-induced chronic asthma without causing side effects
Airway inflammation is the major pathological feature of asthma. Thus, the current therapeutic strategy for asthma is to control inflammation. Limethason, an anti‑inflammation drug, is widely used in rheumatoid arthritis treatment. The aim of the present study was to detect the anti‑inflammatory effect and side effects of limethason on airways that were sensitized with ovalbumin in a murine model of chronic asthma. In the present study, BALB/c mice were sensitized with ovalbumin. Airway hyperresponsiveness was estimated, and hematoxylin and eosin staining, Periodic acid‑Schiff staining and bronchoalveolar lavage were used to detect the effect on chronic asthma. Limethason effectively reduced airway hyperresponsiveness, and inhibited inflammatory cell infiltration and mucus secretion. Bronchoalveolar lavage fluid analysis revealed that limethason suppressed levels of airway eosinophils. In the period of treatment, limethason exhibited no influence on morphology of the femoral head, bone mineral content or bone mineral density, which were detected by histological studies and dual‑energy X‑ray absorptiometry. The index of liver, spleen, kidney, gastrocnemius and brown adipose tissue also demonstrated that limethason had no adverse effects on organs and tissues. The present study revealed that limethason could effectively reduce inflammation in an asthma mouse model without side effects. Therefore, limethason may have therapeutic potential for treating chronic asthma clinically.