Changbin Shao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Changbin Shao has a total of 48 co-authors. Here are the top co-authors by number of shared publications:
- Hualong Yu
School of Computer, Jiangsu University of Science and Technology, Zhenjiang 212003, China
10 shared publications - Chunlong Hu
School of Computer Science, Jiangsu University of Science and Technology, Zhenjiang, China
4 shared publications - Xin Shu
Jiangsu University of Science and Technology
4 shared publications - Bin Qin
School of Computer, Jiangsu University of Science and Technology, Zhenjiang 212003, China
3 shared publications - Zhibin Xie
Jiangsu University of Science and Technology
3 shared publications Hualong Yu
School of Computer, Jiangsu University of Science and Technology, Zhenjiang, China
2 shared publications
Latest Publications
- 25 October 2026
- Published by Springer Nature in Multimedia Systems
- Vol. 32 (10) , 672
High-frequency feature-aware water-related object detection
- 25 October 2026
- Published by Springer Nature in Multimedia Systems
- Vol. 32 (10) , 672
High-frequency feature-aware water-related object detection
- Open Access
Online Adaptive TSK Fuzzy Classifier with Incremental–Decremental Update Scheme for Non-Stationary Data Streams
Takagi–Sugeno–Kang (TSK) fuzzy classifiers have achieved notable success in pattern recognition and data mining due to their combination of approximation capability and model interpretability. However, existing TSK classifiers predominantly rely on batch training, which becomes impractical in streaming environments characterized by continuous sample arrival, limited memory, and evolving data distributions. To address these challenges, this paper proposes the Online Adaptive TSK (OA-TSK) fuzzy classifier with an incremental–decremental update protocol for non-stationary data streams. First, we develop the Decremental Adaptive TSK Fuzzy Classifier (DA-TSK), whose update procedure applies the Sherman–Morrison formula to a symmetric rank-one downdate, enabling precise sample removal while maintaining exact equivalence to batch least-squares retraining. Building upon DA-TSK and our prior Dynamic Incremental Adaptive TSK Fuzzy Classifier (DIA-TSK), we then propose OA-TSK, which integrates incremental and decremental updates within a sliding-window mechanism to achieve continuous adaptation with bounded memory. The key theoretical contribution is that, in exact arithmetic, OA-TSK produces at every time step the same ridge-regularized least-squares solution as batch retraining on the active window. To further handle severe shifts in the input distribution, which can cause fixed rule centers to become poorly aligned with newly arriving samples, we propose Online Adaptive TSK with Antecedent Reclustering (OA-A-TSK), which incorporates an adaptive antecedent reclustering mechanism. Comprehensive experiments on synthetic and real-world data streams demonstrate that OA-TSK achieves classification accuracy comparable to that of ridge-regularized batch retraining while substantially reducing per-sample computational complexity. An analysis of average ranks shows that OA-TSK and OA-A-TSK rank among the top-performing methods across diverse drift scenarios while preserving the intrinsic interpretability of TSK fuzzy systems.
Open AccessOnline Adaptive TSK Fuzzy Classifier with Incremental–Decremental Update Scheme for Non-Stationary Data Streams
Takagi–Sugeno–Kang (TSK) fuzzy classifiers have achieved notable success in pattern recognition and data mining due to their combination of approximation capability and model interpretability. However, existing TSK classifiers predominantly rely on batch training, which becomes impractical in streaming environments characterized by continuous sample arrival, limited memory, and evolving data distributions. To address these challenges, this paper proposes the Online Adaptive TSK (OA-TSK) fuzzy classifier with an incremental–decremental update protocol for non-stationary data streams. First, we develop the Decremental Adaptive TSK Fuzzy Classifier (DA-TSK), whose update procedure applies the Sherman–Morrison formula to a symmetric rank-one downdate, enabling precise sample removal while maintaining exact equivalence to batch least-squares retraining. Building upon DA-TSK and our prior Dynamic Incremental Adaptive TSK Fuzzy Classifier (DIA-TSK), we then propose OA-TSK, which integrates incremental and decremental updates within a sliding-window mechanism to achieve continuous adaptation with bounded memory. The key theoretical contribution is that, in exact arithmetic, OA-TSK produces at every time step the same ridge-regularized least-squares solution as batch retraining on the active window. To further handle severe shifts in the input distribution, which can cause fixed rule centers to become poorly aligned with newly arriving samples, we propose Online Adaptive TSK with Antecedent Reclustering (OA-A-TSK), which incorporates an adaptive antecedent reclustering mechanism. Comprehensive experiments on synthetic and real-world data streams demonstrate that OA-TSK achieves classification accuracy comparable to that of ridge-regularized batch retraining while substantially reducing per-sample computational complexity. An analysis of average ranks shows that OA-TSK and OA-A-TSK rank among the top-performing methods across diverse drift scenarios while preserving the intrinsic interpretability of TSK fuzzy systems.
- 10 October 2026
- Published by Springer Nature in Multimedia Systems
- Vol. 32 (10) , 630
Adaptive gated feature pyramid network for facial age estimation
- 10 October 2026
- Published by Springer Nature in Multimedia Systems
- Vol. 32 (10) , 630
Adaptive gated feature pyramid network for facial age estimation
- 14 August 2026
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 37 (33) , 335401
Enhancing nighttime object detection via joint-teacher: an unsupervised domain adaptation approach
Object detection in nighttime scene presents significant challenges due to domain shifts from daytime environments, primarily caused by insufficient illumination and increased noise. Existing methods often rely on teacher–student frameworks for self-training with pseudo-labels, However, low light and noise increase the difficulty of detection, leading to a significant increase in the positional error between the pseudo-labels and the ground truth. This paper introduces Joint-Teacher, an unsupervised domain adaptation method designed to enhance nighttime object detection. By implementing a student-evaluates-teacher mechanism, we filter reliable pseudo-labels based on regression stability and confidence scores. Additionally, a consistent joint sample strategy is proposed to improve sample allocation accuracy through spatial feature alignment. To simulate real nighttime conditions, a nighttime scene transformation module is designed. Experiments on BDD100K and DWD datasets demonstrate that Joint-Teacher outperforms state-of-the-art methods, achieving an average precision of 49.9 on BDD100K and 47.5 on DWD, validating its effectiveness in complex nighttime scene. The code and pretrained models are available at: https://github.com/Jiaquan789/Joint-Teacher .
- 14 August 2026
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 37 (33) , 335401
Enhancing nighttime object detection via joint-teacher: an unsupervised domain adaptation approach
Object detection in nighttime scene presents significant challenges due to domain shifts from daytime environments, primarily caused by insufficient illumination and increased noise. Existing methods often rely on teacher–student frameworks for self-training with pseudo-labels, However, low light and noise increase the difficulty of detection, leading to a significant increase in the positional error between the pseudo-labels and the ground truth. This paper introduces Joint-Teacher, an unsupervised domain adaptation method designed to enhance nighttime object detection. By implementing a student-evaluates-teacher mechanism, we filter reliable pseudo-labels based on regression stability and confidence scores. Additionally, a consistent joint sample strategy is proposed to improve sample allocation accuracy through spatial feature alignment. To simulate real nighttime conditions, a nighttime scene transformation module is designed. Experiments on BDD100K and DWD datasets demonstrate that Joint-Teacher outperforms state-of-the-art methods, achieving an average precision of 49.9 on BDD100K and 47.5 on DWD, validating its effectiveness in complex nighttime scene. The code and pretrained models are available at: https://github.com/Jiaquan789/Joint-Teacher .
- 19 June 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 349
DSW-Net: A dual-skip connection wavelet network for underwater image enhancement
- 19 June 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 349
DSW-Net: A dual-skip connection wavelet network for underwater image enhancement
- Open Access
Cross-Identity Interaction Transformer for Facial Age Estimation
Despite the remarkable progress being made in the study of human facial age estimation, it is still a challenging problem. The main problem lies in the large intra-age appearance variations among different individuals. Sometimes, these variations can even exceed the inter-age appearance variations of the same individual. To address this problem, we construct a cross-identity image sequence for each query image and reformulate age estimation as a multi-image learning task. This provides a basis for learning common age-related cues across identities. Based on this formulation, we propose the Cross-Identity Interaction Transformer (CIIT) for age estimation. The CIIT first extracts multi-scale aging cues through a cross-scale embedding (CSE) module to preserve age evidence from fine textures to coarse structural changes. Secondly, to progressively enhance facial features and capture shared facial characteristics from cross-identity references, intra-image feature attention (IFA) and prior-guided axial cross-image attention (PG-ACIA) operate alternately within each Transformer block. IFA refines local age-discriminative representations within each image, while PG-ACIA uses multi-scale edge priors to guide cross-image interaction toward age-sensitive regions such as wrinkles. Finally, an anchored regression network (ARN) predicts age through a soft-weighted combination of multiple linear regressors for robust age estimation under diverse facial aging patterns. Experiments on four benchmark datasets, namely MORPH Album II, MegaAge-Asian, FG-NET and Adience, demonstrate that the proposed method achieves superior performance across multiple evaluation metrics, validating the effectiveness of the CIIT in capturing shared facial characteristics.
Open AccessCross-Identity Interaction Transformer for Facial Age Estimation
Despite the remarkable progress being made in the study of human facial age estimation, it is still a challenging problem. The main problem lies in the large intra-age appearance variations among different individuals. Sometimes, these variations can even exceed the inter-age appearance variations of the same individual. To address this problem, we construct a cross-identity image sequence for each query image and reformulate age estimation as a multi-image learning task. This provides a basis for learning common age-related cues across identities. Based on this formulation, we propose the Cross-Identity Interaction Transformer (CIIT) for age estimation. The CIIT first extracts multi-scale aging cues through a cross-scale embedding (CSE) module to preserve age evidence from fine textures to coarse structural changes. Secondly, to progressively enhance facial features and capture shared facial characteristics from cross-identity references, intra-image feature attention (IFA) and prior-guided axial cross-image attention (PG-ACIA) operate alternately within each Transformer block. IFA refines local age-discriminative representations within each image, while PG-ACIA uses multi-scale edge priors to guide cross-image interaction toward age-sensitive regions such as wrinkles. Finally, an anchored regression network (ARN) predicts age through a soft-weighted combination of multiple linear regressors for robust age estimation under diverse facial aging patterns. Experiments on four benchmark datasets, namely MORPH Album II, MegaAge-Asian, FG-NET and Adience, demonstrate that the proposed method achieves superior performance across multiple evaluation metrics, validating the effectiveness of the CIIT in capturing shared facial characteristics.
- Open Access
- 8 May 2026
- Published by MDPI AG in Information
- Vol. 17 (5) , 455
ACVM: An Adaptive Combination Validation Mechanism for Long-Tailed Image Recognition
In real-world scenarios, large-scale datasets often exhibit a long-tailed data distribution. Training deep neural networks on such data typically leads to a bias towards head classes. Existing studies have demonstrated that the reweighting strategy is an effective means to alleviate the long-tailed issue. Recent studies suggest that incorporating class difficulty into reweighting can yield superior results. However, the method of quantifying class difficulty by an independent validation set has shown limitations in practical applications, i.e., wasting training samples and inaccurate estimations. To address this issue, this study proposes a novel model based on K-fold cross-validation, called the adaptive combination validation model, which contains two main innovations: first, both class and sample difficulty are quantified by using a more comprehensive and authentic estimation strategy, i.e., K-fold cross-validation, to obtain accurate and robust estimations; second, we extract the prediction probability distributions of samples, which reflect sample difficulty, from different model branches and design a distribution-harmonized loss to simultaneously focus on the effects of reweighted and original distributions. Extensive experiments on several popular long-tailed image recognition datasets (CIFAR10-LT and CIFAR100-LT, with several varying imbalance rates, and ImageNet-LT) demonstrate that the proposed method can effectively alleviate the long-tailed issue and achieve state-of-the-art performance on most datasets.
- 8 May 2026
- Published by MDPI AG in Information
- Vol. 17 (5) , 455
Open AccessACVM: An Adaptive Combination Validation Mechanism for Long-Tailed Image Recognition
In real-world scenarios, large-scale datasets often exhibit a long-tailed data distribution. Training deep neural networks on such data typically leads to a bias towards head classes. Existing studies have demonstrated that the reweighting strategy is an effective means to alleviate the long-tailed issue. Recent studies suggest that incorporating class difficulty into reweighting can yield superior results. However, the method of quantifying class difficulty by an independent validation set has shown limitations in practical applications, i.e., wasting training samples and inaccurate estimations. To address this issue, this study proposes a novel model based on K-fold cross-validation, called the adaptive combination validation model, which contains two main innovations: first, both class and sample difficulty are quantified by using a more comprehensive and authentic estimation strategy, i.e., K-fold cross-validation, to obtain accurate and robust estimations; second, we extract the prediction probability distributions of samples, which reflect sample difficulty, from different model branches and design a distribution-harmonized loss to simultaneously focus on the effects of reweighted and original distributions. Extensive experiments on several popular long-tailed image recognition datasets (CIFAR10-LT and CIFAR100-LT, with several varying imbalance rates, and ImageNet-LT) demonstrate that the proposed method can effectively alleviate the long-tailed issue and achieve state-of-the-art performance on most datasets.
- Open Access
Entropy Subtraction-Supported Residual-Diffusion Framework for Image Super-Resolution
Diffusion probabilistic models have demonstrated remarkable superiority in SISR. Yet, their multi-step denoising mechanism incurs prohibitive computational overhead, which severely limits real-world deployment. To address this issue, we propose an Entropy Subtraction-Supported Diffusion Denoising framework for image Reconstruction (ESRDF). The core idea is to shift part of the SR burden from the diffusion model to an image Decoder, with a key focus on recovering the symmetric structural correspondence between LR and HR images that is often degraded during downsampling. Specifically, ESRDF’s main branch employs a CNN that performs one-step feature reconstruction, supervised by a novel entropy-matching loss in addition to the conventional reconstruction loss. This loss adopts a patch-wise entropy matching strategy that enforces regional consistency between the True and the predicted images. Building on L1’s focus on pixel-level details and perceptual loss’s grasp of global semantics, region-wise entropy measurement further completes the global alignment of intra-region information structures. Under this framework, the main branch delivers coarse low-frequency content, drastically reducing the workload of the diffusion branch, which now only needs to sparsely refine high-frequency details. Experimental results on multiple benchmark datasets demonstrate that ESRDF achieves shorter model convergence times and higher generation quality with fewer denoising steps, outperforming previous diffusion-based image reconstruction methods.
Open AccessEntropy Subtraction-Supported Residual-Diffusion Framework for Image Super-Resolution
Diffusion probabilistic models have demonstrated remarkable superiority in SISR. Yet, their multi-step denoising mechanism incurs prohibitive computational overhead, which severely limits real-world deployment. To address this issue, we propose an Entropy Subtraction-Supported Diffusion Denoising framework for image Reconstruction (ESRDF). The core idea is to shift part of the SR burden from the diffusion model to an image Decoder, with a key focus on recovering the symmetric structural correspondence between LR and HR images that is often degraded during downsampling. Specifically, ESRDF’s main branch employs a CNN that performs one-step feature reconstruction, supervised by a novel entropy-matching loss in addition to the conventional reconstruction loss. This loss adopts a patch-wise entropy matching strategy that enforces regional consistency between the True and the predicted images. Building on L1’s focus on pixel-level details and perceptual loss’s grasp of global semantics, region-wise entropy measurement further completes the global alignment of intra-region information structures. Under this framework, the main branch delivers coarse low-frequency content, drastically reducing the workload of the diffusion branch, which now only needs to sparsely refine high-frequency details. Experimental results on multiple benchmark datasets demonstrate that ESRDF achieves shorter model convergence times and higher generation quality with fewer denoising steps, outperforming previous diffusion-based image reconstruction methods.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 33, 2345-2349
High-Frequency Information Supported Domain Adaptation for Cross-Domain Object Detection
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 33, 2345-2349
High-Frequency Information Supported Domain Adaptation for Cross-Domain Object Detection
- 22 December 2025
- Published by Elsevier in Applied Soft Computing
- Vol. 189
Degradation-aware deep unfolding network with multi-loss optimization for underwater image enhancement
- 22 December 2025
- Published by Elsevier in Applied Soft Computing
- Vol. 189
Degradation-aware deep unfolding network with multi-loss optimization for underwater image enhancement