Chun Yang
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
Research Subjects
Research Interests
Chun Yang has a total of 48 co-authors. Here are the top co-authors by number of shared publications:
- Xu-Cheng Yin
University of Science and Technology Beijing, Beijing, China
28 shared publications - Chang Liu
Luleå University of Technology
15 shared publications - Xiaobin Zhu
University of Science and Technology Beijing, Beijing, China
10 shared publications - Jingyan Qin
School of Intelligence Science and Technology University of Science and Technology Beijing
4 shared publications - Shi-Xue Zhang
School of Computer and Communication Engineering, University of Science and Technology Beijing (USTB), Beijing, China
4 shared publications - Fang Zhou
University of Science and Technology Beijing
3 shared publications
Latest Publications
- Open Access
- 27 September 2026
- Published by MDPI AG in Electronics
- Vol. 15 (19) , 4449
High-Resolution Perception at Scale: Benchmarking Dense Geometric Models via Resolution-Calibrated Corruptions
While recent advancements in large-scale pretraining have significantly improved dense geometric models, these architectures often struggle to adapt to 4K-resolution inputs because of their rigidity. Existing benchmarks mostly provide low-resolution samples and lack quantitative evaluations on high-resolution corrupted inputs. We address these gaps in two ways. First, we propose the Resolution-Calibrated Corruptions (RCC) framework to dynamically scale perturbations. Second, we introduce a benchmark with over 22,000 samples to evaluate optical flow and depth estimation architectures at 1K, 2K, and 4K resolutions. Our evaluation reveals a dual challenge: natively processing 4K inputs exposes architectural rigidities, while workarounds like input downsampling or spatial tiling may introduce spatial-aliasing artifacts or lose global context. Experiments show that traditional optical flow architectures degrade considerably under dense weather occlusions at standard resolutions. When natively processing 4K inputs, spatial redundancy mitigates these sparse occlusions, but—under our simulated physical sensor constraints—exposes a noticeable vulnerability to high-frequency sensor noise and yields higher absolute errors. Consequently, while models leveraging geometric priors remain noticeably more robust to noise and scale changes, input downsampling remains the most pragmatic choice for overall accuracy, despite potentially introducing spatial aliasing artifacts.
- 27 September 2026
- Published by MDPI AG in Electronics
- Vol. 15 (19) , 4449
Open AccessHigh-Resolution Perception at Scale: Benchmarking Dense Geometric Models via Resolution-Calibrated Corruptions
While recent advancements in large-scale pretraining have significantly improved dense geometric models, these architectures often struggle to adapt to 4K-resolution inputs because of their rigidity. Existing benchmarks mostly provide low-resolution samples and lack quantitative evaluations on high-resolution corrupted inputs. We address these gaps in two ways. First, we propose the Resolution-Calibrated Corruptions (RCC) framework to dynamically scale perturbations. Second, we introduce a benchmark with over 22,000 samples to evaluate optical flow and depth estimation architectures at 1K, 2K, and 4K resolutions. Our evaluation reveals a dual challenge: natively processing 4K inputs exposes architectural rigidities, while workarounds like input downsampling or spatial tiling may introduce spatial-aliasing artifacts or lose global context. Experiments show that traditional optical flow architectures degrade considerably under dense weather occlusions at standard resolutions. When natively processing 4K inputs, spatial redundancy mitigates these sparse occlusions, but—under our simulated physical sensor constraints—exposes a noticeable vulnerability to high-frequency sensor noise and yields higher absolute errors. Consequently, while models leveraging geometric priors remain noticeably more robust to noise and scale changes, input downsampling remains the most pragmatic choice for overall accuracy, despite potentially introducing spatial aliasing artifacts.
- Open Access
- 15 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 2826-2832
Beyond Optical Flow: Latent Micro-Motion as Visual Evidence in UAV Video
- 15 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 2826-2832
Open AccessBeyond Optical Flow: Latent Micro-Motion as Visual Evidence in UAV Video
- 8 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (8) , 4868-4882
Amplitude–Phase Reconstruction for Non-Stationary Time-Series Forecasting
Real-world time-series can be decomposed into multiple interacting frequency components whose amplitudes and phases co-evolve over time. Such coupled dynamics can deform spectral trajectories, leading to pronounced non-stationarity in frequency-domain representations and substantial degradation in long-horizon forecasting performance. However, most existing frequency-domain forecasting methods do not explicitly model amplitude–phase interactions and often implicitly assume globally consistent spectral structures, which limits their ability to capture drifting spectra under non-stationary dynamics. To address this challenge, we propose a novel Amplitude–Phase Reconstruction Network (APRNet), a frequency-domain time-series forecasting framework that models amplitude–phase interactions from temporal and channel-wise perspectives. Specifically, we propose an innovative Amplitude-Phase Global Correlation (APGC) module to capture spectrum-wide amplitude-phase dependencies and derive frequency-wise calibration factors under non-stationary dynamics. The calibrated spectral representations are further reconstructed into the time domain, forming a closed time-frequency reconstruction loop that suppresses spectral drift and mitigates non-stationarity in temporal features. In addition, we propose a novel Discrete Kolmogorov–Arnold Network (D-KAN) to further enhance fine-grained amplitude–phase modeling. By combining discretized nonlinear activations with locally supported B-spline basis functions, D-KAN enables frequency-adaptive piecewise nonlinear modeling, improving sensitivity to localized spectral variations and enhancing high-frequency expressiveness. Extensive experiments verify the superior performance of our APRNet. Our codes are available at: https://github.com/LH325/APRNet.
- 8 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (8) , 4868-4882
Amplitude–Phase Reconstruction for Non-Stationary Time-Series Forecasting
Real-world time-series can be decomposed into multiple interacting frequency components whose amplitudes and phases co-evolve over time. Such coupled dynamics can deform spectral trajectories, leading to pronounced non-stationarity in frequency-domain representations and substantial degradation in long-horizon forecasting performance. However, most existing frequency-domain forecasting methods do not explicitly model amplitude–phase interactions and often implicitly assume globally consistent spectral structures, which limits their ability to capture drifting spectra under non-stationary dynamics. To address this challenge, we propose a novel Amplitude–Phase Reconstruction Network (APRNet), a frequency-domain time-series forecasting framework that models amplitude–phase interactions from temporal and channel-wise perspectives. Specifically, we propose an innovative Amplitude-Phase Global Correlation (APGC) module to capture spectrum-wide amplitude-phase dependencies and derive frequency-wise calibration factors under non-stationary dynamics. The calibrated spectral representations are further reconstructed into the time domain, forming a closed time-frequency reconstruction loop that suppresses spectral drift and mitigates non-stationarity in temporal features. In addition, we propose a novel Discrete Kolmogorov–Arnold Network (D-KAN) to further enhance fine-grained amplitude–phase modeling. By combining discretized nonlinear activations with locally supported B-spline basis functions, D-KAN enables frequency-adaptive piecewise nonlinear modeling, improving sensitivity to localized spectral variations and enhancing high-frequency expressiveness. Extensive experiments verify the superior performance of our APRNet. Our codes are available at: https://github.com/LH325/APRNet.
- 2 February 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 312
Local attention alignment fusion network for domain adaptive water body segmentation
- 2 February 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 312
Local attention alignment fusion network for domain adaptive water body segmentation
- 13 September 2025
- Published by Springer Nature
- p. 351-369
T-LLaVA: An Effective Saliency-Aware Slicing Strategy for Text Recognition
- 13 September 2025
- Published by Springer Nature
- p. 351-369
T-LLaVA: An Effective Saliency-Aware Slicing Strategy for Text Recognition
- Open Access
Stationarity Exploration for Multivariate Time Series Forecasting
Deep learning-based time series forecasting has found widespread applications. Recently, converting time series data into the frequency domain for forecasting has become popular for accurately exploring periodic patterns. However, existing methods often cannot effectively explore stationary information from complex intertwined frequency components. In this paper, we propose a simple yet effective Amplitude-Phase Reconstruct Network (APRNet) that models the inter-relationships of amplitude and phase, which prevents the amplitude and phase from being constrained by different physical quantities, thereby decoupling the distinct characteristics of signals for capturing stationary information. Specifically, we represent the multivariate time series input across sequence and channel dimensions, highlighting the correlation between amplitude and phase at multiple interaction frequencies. We propose a novel Kolmogorov-Arnold-Network-based Local Correlation (KLC) module to adaptively fit local functions using univariate functions, enabling more flexible characterization of stationary features across different amplitudes and phases. This significantly enhances the model's capability to capture time-varying patterns. Extensive experiments demonstrate the superiority of our APRNet against the state-of-the-arts (SOTAs).
Open AccessStationarity Exploration for Multivariate Time Series Forecasting
Deep learning-based time series forecasting has found widespread applications. Recently, converting time series data into the frequency domain for forecasting has become popular for accurately exploring periodic patterns. However, existing methods often cannot effectively explore stationary information from complex intertwined frequency components. In this paper, we propose a simple yet effective Amplitude-Phase Reconstruct Network (APRNet) that models the inter-relationships of amplitude and phase, which prevents the amplitude and phase from being constrained by different physical quantities, thereby decoupling the distinct characteristics of signals for capturing stationary information. Specifically, we represent the multivariate time series input across sequence and channel dimensions, highlighting the correlation between amplitude and phase at multiple interaction frequencies. We propose a novel Kolmogorov-Arnold-Network-based Local Correlation (KLC) module to adaptively fit local functions using univariate functions, enabling more flexible characterization of stationary features across different amplitudes and phases. This significantly enhances the model's capability to capture time-varying patterns. Extensive experiments demonstrate the superiority of our APRNet against the state-of-the-arts (SOTAs).
- 11 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 33 (10577149) , 6497-6507
CFOR: Character-First Open-Set Text Recognition via Context-Free Learning
The open-set text recognition task is a generalized form of the (close-set) text recognition task, where the model is further challenged to spot and incrementally recognize novel characters in new languages, which also indicate a shift in the language model. In this work, we propose to alleviate the confounding effect such biases under an open-set setup by learning individual character representations out of their context.We propose a Character-First Open-Set Text Recognition framework that treats individual characters as first-class citizens by cotraining the feature extractor with two context-free learning tasks. First, a Context Isolation Learning task is proposed to wipe the context for each character, utilizing a character mask learned in a weak supervision manner. Second, the framework adopts an Individual Character Learning task, which is a single-character classification task with synthetic samples. Our framework can reliably spot unseen characters in Japanese with F1-score over 64%, and can adapt to recognize unseen characters in Japanese, Korean, Greek, and other languages without retraining. The framework also shows decent many-shot performance on close-set text benchmarks with 91.5% line accuracy on IIIT5k and a speed of over 69 FPS single-batched, making it a feasible universal lightweight OCR solution.
- 11 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 33 (10577149) , 6497-6507
CFOR: Character-First Open-Set Text Recognition via Context-Free Learning
The open-set text recognition task is a generalized form of the (close-set) text recognition task, where the model is further challenged to spot and incrementally recognize novel characters in new languages, which also indicate a shift in the language model. In this work, we propose to alleviate the confounding effect such biases under an open-set setup by learning individual character representations out of their context.We propose a Character-First Open-Set Text Recognition framework that treats individual characters as first-class citizens by cotraining the feature extractor with two context-free learning tasks. First, a Context Isolation Learning task is proposed to wipe the context for each character, utilizing a character mask learned in a weak supervision manner. Second, the framework adopts an Individual Character Learning task, which is a single-character classification task with synthetic samples. Our framework can reliably spot unseen characters in Japanese with F1-score over 64%, and can adapt to recognize unseen characters in Japanese, Korean, Greek, and other languages without retraining. The framework also shows decent many-shot performance on close-set text benchmarks with 91.5% line accuracy on IIIT5k and a speed of over 69 FPS single-batched, making it a feasible universal lightweight OCR solution.
- Open Access
SAN: Structure-Aware Network for Complex and Long-tailed Chinese Text Recognition
In text recognition, complex glyphs and tail classes have always been factors affecting model performance. Specifically for Chinese text recognition, the lack of shape-awareness can lead to confusion among close complex characters. Since such characters are often tail classes that appear less frequently in the training-set, making it harder for the model to capture its shape information. Hence in this work, we propose a structure-aware network utilizing the hierarchical composition information to improve the recognition performance of complex characters. Implementation-wise, we first propose an auxiliary radical branch and integrate it into the base recognition network as a regularization term, which distills hierarchical composition information into the feature extractor. A Tree-Similarity-based weighting mechanism is then proposed to further utilize the depth information in the hierarchical representation. Experiments demonstrate that the proposed approach can significantly improve the performances of complex characters and tail characters, yielding a better overall performance. Code is available at https://github.com/Levi-ZJY/SAN.
Open AccessSAN: Structure-Aware Network for Complex and Long-tailed Chinese Text Recognition
In text recognition, complex glyphs and tail classes have always been factors affecting model performance. Specifically for Chinese text recognition, the lack of shape-awareness can lead to confusion among close complex characters. Since such characters are often tail classes that appear less frequently in the training-set, making it harder for the model to capture its shape information. Hence in this work, we propose a structure-aware network utilizing the hierarchical composition information to improve the recognition performance of complex characters. Implementation-wise, we first propose an auxiliary radical branch and integrate it into the base recognition network as a regularization term, which distills hierarchical composition information into the feature extractor. A Tree-Similarity-based weighting mechanism is then proposed to further utilize the depth information in the hierarchical representation. Experiments demonstrate that the proposed approach can significantly improve the performances of complex characters and tail characters, yielding a better overall performance. Code is available at https://github.com/Levi-ZJY/SAN.
- Open Access
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 1701-1710
Exploring Stable Meta-Optimization Patterns via Differentiable Reinforcement Learning for Few-Shot Classification
Existing few-shot learning methods generally focus on designing exquisite structures of meta-learners for learning task-specific prior to improve the discriminative ability of global embeddings. However, they often ignore the importance of learning stability in meta-training, making it difficult to obtain a relatively optimal model. From this key observation, we propose an innovative generic differentiable Reinforcement Learning (RL) strategy for few-shot classification. It aims to explore stable meta-optimization patterns in meta-training by learning generalizable optimizations for producing task-adaptive embeddings. Accordingly, our differentiable RL strategy models the embedding procedure of feature transformation layers in meta-learner to optimize the gradient flow implicitly. Also, we propose a memory module to associate historical and current task states and actions for exploring inter-task similarity. Notably, our RL-based strategy can be easily extended to various backbones. In addition, we propose a novel task state encoder to encode task representation, which fully explores inner-task similarities between support set and query set. Extensive experiments verify that our approach can improve the performance of different backbones and achieve promising results against state-of-the-art methods in few-shot classification.
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 1701-1710
Open AccessExploring Stable Meta-Optimization Patterns via Differentiable Reinforcement Learning for Few-Shot Classification
Existing few-shot learning methods generally focus on designing exquisite structures of meta-learners for learning task-specific prior to improve the discriminative ability of global embeddings. However, they often ignore the importance of learning stability in meta-training, making it difficult to obtain a relatively optimal model. From this key observation, we propose an innovative generic differentiable Reinforcement Learning (RL) strategy for few-shot classification. It aims to explore stable meta-optimization patterns in meta-training by learning generalizable optimizations for producing task-adaptive embeddings. Accordingly, our differentiable RL strategy models the embedding procedure of feature transformation layers in meta-learner to optimize the gradient flow implicitly. Also, we propose a memory module to associate historical and current task states and actions for exploring inter-task similarity. Notably, our RL-based strategy can be easily extended to various backbones. In addition, we propose a novel task state encoder to encode task representation, which fully explores inner-task similarities between support set and query set. Extensive experiments verify that our approach can improve the performance of different backbones and achieve promising results against state-of-the-art methods in few-shot classification.
- Open Access
- 10 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 70-79
Transformer-based Reasoning for Learning Evolutionary Chain of Events on Temporal Knowledge Graph
Temporal Knowledge Graph (TKG) reasoning often involves completing missing factual elements along the timeline. Although existing methods can learn good embeddings for each factual element in quadruples by integrating temporal information, they often fail to infer the evolution of temporal facts. This is mainly because of (1) insufficiently exploring the internal structure and semantic relationships within individual quadruples and (2) inadequately learning a unified representation of the contextual and temporal correlations among different quadruples. To overcome these limitations, we propose a novel Transformer-based reasoning model (dubbed ECEformer) for TKG to learn the Evolutionary Chain of Events (ECE). Specifically, we unfold the neighborhood subgraph of an entity node in chronological order, forming an evolutionary chain of events as the input for our model. Subsequently, we utilize a Transformer encoder to learn the embeddings of intra-quadruples for ECE. We then craft a mixed-context reasoning module based on the multi-layer perceptron (MLP) to learn the unified representations of inter-quadruples for ECE while accomplishing temporal knowledge reasoning. In addition, to enhance the timeliness of the events, we devise an additional time prediction task to complete effective temporal information within the learned unified representation. Extensive experiments on six benchmark datasets verify the state-of-the-art performance and the effectiveness of our method.
- 10 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 70-79
Open AccessTransformer-based Reasoning for Learning Evolutionary Chain of Events on Temporal Knowledge Graph
Temporal Knowledge Graph (TKG) reasoning often involves completing missing factual elements along the timeline. Although existing methods can learn good embeddings for each factual element in quadruples by integrating temporal information, they often fail to infer the evolution of temporal facts. This is mainly because of (1) insufficiently exploring the internal structure and semantic relationships within individual quadruples and (2) inadequately learning a unified representation of the contextual and temporal correlations among different quadruples. To overcome these limitations, we propose a novel Transformer-based reasoning model (dubbed ECEformer) for TKG to learn the Evolutionary Chain of Events (ECE). Specifically, we unfold the neighborhood subgraph of an entity node in chronological order, forming an evolutionary chain of events as the input for our model. Subsequently, we utilize a Transformer encoder to learn the embeddings of intra-quadruples for ECE. We then craft a mixed-context reasoning module based on the multi-layer perceptron (MLP) to learn the unified representations of inter-quadruples for ECE while accomplishing temporal knowledge reasoning. In addition, to enhance the timeliness of the events, we devise an additional time prediction task to complete effective temporal information within the learned unified representation. Extensive experiments on six benchmark datasets verify the state-of-the-art performance and the effectiveness of our method.