Yin Tian
Research Subjects
Research Interests
Yin Tian has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Xiaoyu Xian
CRRC Academy Company Ltd., Beijing, China
3 shared publications - Daxin Tian
Beijing Advanced Innovation Center for Big Data and Brain Computing, Beijing Key Laboratory for Cooperative Vehicle Infrastructure Systems and Safety Control, School of Transportation Science and Engineering, Beihang University, Beijing, China
2 shared publications - Haichuan Tang
CRRC Industrial Academy Co., Ltd., Beijing 100071, China
2 shared publications - Yukai Shi
Guangdong University of Technology
2 shared publications Boshi Li
Beijing Aerospace Kangtuo Information Technology Co., Ltd
1 shared publication- Chenhui Shao
Department of Mechanical Science and Engineering University of Illinois at Urbana-Champaign 61801 Urbana IL USA
1 shared publication
Latest Publications
- 6 March 2026
- Published by Elsevier in Pattern Recognition
- Vol. 178
Physics-informed visual-inertial mamba for robust train localization in harsh conditions
- 6 March 2026
- Published by Elsevier in Pattern Recognition
- Vol. 178
Physics-informed visual-inertial mamba for robust train localization in harsh conditions
- Open Access
CrossFuse: Learning Infrared and Visible Image Fusion by Cross-Sensor Top-K Vision Alignment and Beyond
Infrared and visible image fusion (IVIF) is increasingly applied in critical fields such as video surveillance and autonomous driving systems. Significant progress has been made in deep learning-based fusion methods. However, these models frequently encounter out-of-distribution (OOD) scenes in real-world applications, which severely impact their performance and reliability. Therefore, addressing the challenge of OOD data is crucial for the safe deployment of these models in open-world environments. Unlike existing research, our focus is on the challenges posed by OOD data in real-world applications and on enhancing the robustness and generalization of models. In this paper, we propose an infrared-visible fusion framework based on Multi-View Augmentation. For external data augmentation, Top-k Selective Vision Alignment is employed to mitigate distribution shifts between datasets by performing RGB-wise transformations on visible images. This strategy effectively introduces augmented samples, enhancing the adaptability of the model to complex real-world scenarios. Additionally, for internal data augmentation, self-supervised learning is established using Weak-Aggressive Augmentation. This enables the model to learn more robust and general feature representations during the fusion process, thereby improving robustness and generalization. Extensive experiments demonstrate that the proposed method exhibits superior performance and robustness across various conditions and environments. Our approach significantly enhances the reliability and stability of IVIF tasks in practical applications.
Open AccessCrossFuse: Learning Infrared and Visible Image Fusion by Cross-Sensor Top-K Vision Alignment and Beyond
Infrared and visible image fusion (IVIF) is increasingly applied in critical fields such as video surveillance and autonomous driving systems. Significant progress has been made in deep learning-based fusion methods. However, these models frequently encounter out-of-distribution (OOD) scenes in real-world applications, which severely impact their performance and reliability. Therefore, addressing the challenge of OOD data is crucial for the safe deployment of these models in open-world environments. Unlike existing research, our focus is on the challenges posed by OOD data in real-world applications and on enhancing the robustness and generalization of models. In this paper, we propose an infrared-visible fusion framework based on Multi-View Augmentation. For external data augmentation, Top-k Selective Vision Alignment is employed to mitigate distribution shifts between datasets by performing RGB-wise transformations on visible images. This strategy effectively introduces augmented samples, enhancing the adaptability of the model to complex real-world scenarios. Additionally, for internal data augmentation, self-supervised learning is established using Weak-Aggressive Augmentation. This enables the model to learn more robust and general feature representations during the fusion process, thereby improving robustness and generalization. Extensive experiments demonstrate that the proposed method exhibits superior performance and robustness across various conditions and environments. Our approach significantly enhances the reliability and stability of IVIF tasks in practical applications.
- 12 February 2024
- Published by Elsevier in Expert Systems with Applications
- Vol. 248
CROSE: Low-light enhancement by CROss-SEnsor interaction for nighttime driving scenes
- 12 February 2024
- Published by Elsevier in Expert Systems with Applications
- Vol. 248
CROSE: Low-light enhancement by CROss-SEnsor interaction for nighttime driving scenes
- 9 January 2024
- Published by Springer Nature
- p. 183-192
Detecting Anomalies in Traction Motor Bearing Using Multi-scale Segmented Autoregressive Network
- 9 January 2024
- Published by Springer Nature
- p. 183-192
Detecting Anomalies in Traction Motor Bearing Using Multi-scale Segmented Autoregressive Network
- Open Access
- 21 October 2022
- Published by Association for Computing Machinery (ACM)
- p. 1771-1778
Fast Pedestrian Detection Algorithm Based on Improved YOLOv3
Aiming at the problems of fast-moving speed, easy occlusion, and complex background of pedestrians in traffic scenes, a fast pedestrian detection algorithm based on improved YOLOv3 is proposed. First, choose the efficient lightweight network ShuffleNetv2 to replace the original backbone network Darknet-53 to reduce the model complexity and improve the detection speed. Second, a reverse residual structure is introduced in the detection network layer to enhance the expressiveness of features. Third, a coordinate attention mechanism is introduced to suppress useless information and enhance the network's ability to focus on key features. Fourth, the spatial pyramid pooling structure is introduced to realize multi-scale feature fusion of the network and improve the detection accuracy of small objects. The experimental results show that compared with YOLOv3, the improved YOLOv3 proposed in this paper can improve the detection accuracy and detection speed by 0.7% and 53.8% respectively, which is more conducive to the rapid detection of pedestrians.
- 21 October 2022
- Published by Association for Computing Machinery (ACM)
- p. 1771-1778
Open AccessFast Pedestrian Detection Algorithm Based on Improved YOLOv3
Aiming at the problems of fast-moving speed, easy occlusion, and complex background of pedestrians in traffic scenes, a fast pedestrian detection algorithm based on improved YOLOv3 is proposed. First, choose the efficient lightweight network ShuffleNetv2 to replace the original backbone network Darknet-53 to reduce the model complexity and improve the detection speed. Second, a reverse residual structure is introduced in the detection network layer to enhance the expressiveness of features. Third, a coordinate attention mechanism is introduced to suppress useless information and enhance the network's ability to focus on key features. Fourth, the spatial pyramid pooling structure is introduced to realize multi-scale feature fusion of the network and improve the detection accuracy of small objects. The experimental results show that compared with YOLOv3, the improved YOLOv3 proposed in this paper can improve the detection accuracy and detection speed by 0.7% and 53.8% respectively, which is more conducive to the rapid detection of pedestrians.
- Open Access
- 29 June 2022
- Published by PHM Society in Phm Society European Conference
- Vol. 7 (1) , 563-570
Prediction of Production Line Status for Printed Circuit Boards
This paper focuses on the problem of predicting production line status for Printed Circuit Boards (PCBs). The problem contains three prediction tasks regarding PCB producing process. Firstly, data exploration is carried out and it reveals several data challenges, including data imbalance, data noise, small sample size, and component difference. To predict production line status for components of PCBs using records of inspection on pins, we proposed two possible feature extraction methods to compress the pin-level data into component-level. A statistical feature extraction method, which retrieves descriptive statistics such as mean, standard deviation, maximum, and minimum of pins on the same component, is applied to Task 1, while a PinNumber-based feature extraction method, which keep original values for 2-pin components, is applied to Task3. In addition, a neural-net model with feeding imbalance control is established for Task 1. and a random forests model is applied for both Task 2 and Task 3. Moreover, a threshold moving technique is proposed to optimize the threshold selection. Finally, the result shows that our models achieved f1-scores of 0.44, 0.54, and 0.71 on the test set for the three tasks, respectively.
- 29 June 2022
- Published by PHM Society in Phm Society European Conference
- Vol. 7 (1) , 563-570
Open AccessPrediction of Production Line Status for Printed Circuit Boards
This paper focuses on the problem of predicting production line status for Printed Circuit Boards (PCBs). The problem contains three prediction tasks regarding PCB producing process. Firstly, data exploration is carried out and it reveals several data challenges, including data imbalance, data noise, small sample size, and component difference. To predict production line status for components of PCBs using records of inspection on pins, we proposed two possible feature extraction methods to compress the pin-level data into component-level. A statistical feature extraction method, which retrieves descriptive statistics such as mean, standard deviation, maximum, and minimum of pins on the same component, is applied to Task 1, while a PinNumber-based feature extraction method, which keep original values for 2-pin components, is applied to Task3. In addition, a neural-net model with feeding imbalance control is established for Task 1. and a random forests model is applied for both Task 2 and Task 3. Moreover, a threshold moving technique is proposed to optimize the threshold selection. Finally, the result shows that our models achieved f1-scores of 0.44, 0.54, and 0.71 on the test set for the three tasks, respectively.
- Open Access
- 8 July 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 25 (5) , 2167-2176
Robust Deep Learning-Based Diagnosis of Mixed Faults in Rotating Machinery
Fault diagnosis for rolling elements in rotating machinery persistently receives high research interest due to the said machinery's prevalence in a broad range of applications. State-of-the-art methods in such setups focus on effective identification of faults that usually involve a single component while rejecting noise from limited sources. This paper studies the data-based diagnosis of mixed faults coming from multiple components with an emphasis on model robustness against a wide spectrum of external perturbation. A dataset is collected on a rotor and bearing system by varying the levels and types of faults in both the rotor and bearing, which results in 48 machine health conditions. A duplet classifier is developed by combining two 1-D convolutional neural networks (CNNs) that are responsible for the diagnosis of rotor and bearing faults, respectively. Experimental results show that the proposed classifier can reliably identify the onset and nature of mixed faults. In addition, one-vs-all classifiers are built using the features generated by the developed 1-D CNNs as predictors to recognize previously unlearned fault types. The effectiveness of such classifiers is demonstrated using data collected from four new fault types. Finally, the robustness and ability to reject external perturbation of the duplet classification model are analyzed using kernel density estimation (KDE).
- 8 July 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 25 (5) , 2167-2176
Open AccessRobust Deep Learning-Based Diagnosis of Mixed Faults in Rotating Machinery
Fault diagnosis for rolling elements in rotating machinery persistently receives high research interest due to the said machinery's prevalence in a broad range of applications. State-of-the-art methods in such setups focus on effective identification of faults that usually involve a single component while rejecting noise from limited sources. This paper studies the data-based diagnosis of mixed faults coming from multiple components with an emphasis on model robustness against a wide spectrum of external perturbation. A dataset is collected on a rotor and bearing system by varying the levels and types of faults in both the rotor and bearing, which results in 48 machine health conditions. A duplet classifier is developed by combining two 1-D convolutional neural networks (CNNs) that are responsible for the diagnosis of rotor and bearing faults, respectively. Experimental results show that the proposed classifier can reliably identify the onset and nature of mixed faults. In addition, one-vs-all classifiers are built using the features generated by the developed 1-D CNNs as predictors to recognize previously unlearned fault types. The effectiveness of such classifiers is demonstrated using data collected from four new fault types. Finally, the robustness and ability to reject external perturbation of the duplet classification model are analyzed using kernel density estimation (KDE).