Teng Long
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Teng Long has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
- Qi Gao
School of Information Engineering, China University of Geosciences (Beijing), Beijing 100083, China
3 shared publications - Zhangbing Zhou
School of Information Engineering, and Frontiers Science Center for Deep-time Digital Earth, China University of Geosciences (Beijing), Beijing, China
3 shared publications Chao Wang
2 shared publicationsQing Wang
2 shared publicationsXingtao Ren
2 shared publications- Zhiwu Xu
College of Computer Science and Software Engineering, Shenzhen University, Shenzhen, China
2 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Software Engineering
- Vol. 52 (11) , 3480-3495
SAFTerm : A Structure-Aware Framework for Termination Analysis via Neural Networks
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Software Engineering
- Vol. 52 (11) , 3480-3495
SAFTerm : A Structure-Aware Framework for Termination Analysis via Neural Networks
- Open Access
- 7 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 1664-1676
Localizing Data Annotation Faults in Semantic Segmentation Systems
- 7 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 1664-1676
Open AccessLocalizing Data Annotation Faults in Semantic Segmentation Systems
- Open Access
- 22 September 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (19) , 9409
A YOLOv8-Based Model for Small-Target Road Defect Detection
Accurate detection of small road defects in complex backgrounds remains challenging due to limited target features and substantial background interference. To improve detection accuracy and model robustness, this study proposes an enhanced road defect detection framework integrating a multi-attention detection head, a structurally re-parameterized Visual Geometry Group (VGG) architecture, and Masked Generative Distillation (MGD). The multi-attention detection head enlarges the receptive field and enhances the model’s ability to capture and distinguish small-target features in complex scenes. The structurally re-parameterized VGG architecture simplifies the network structure while improving detection accuracy. In addition, MGD is employed to transfer discriminative knowledge from a teacher model to a student model, thereby strengthening feature representation and generalization capability. Experiments on the officially released Road Defect Detection 2022 (RDD2022) dataset show that the proposed model improves mean average precision (mAP) by 3.4% over the baseline model. On the Street View Image Dataset for Automated Road Damage Detection (SVRDD), the model achieves a 5.9% mAP improvement. These results demonstrate that the proposed framework effectively improves road defect detection performance across both datasets and provides a feasible approach for robust detection in complex road environments.
- 22 September 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (19) , 9409
Open AccessA YOLOv8-Based Model for Small-Target Road Defect Detection
Accurate detection of small road defects in complex backgrounds remains challenging due to limited target features and substantial background interference. To improve detection accuracy and model robustness, this study proposes an enhanced road defect detection framework integrating a multi-attention detection head, a structurally re-parameterized Visual Geometry Group (VGG) architecture, and Masked Generative Distillation (MGD). The multi-attention detection head enlarges the receptive field and enhances the model’s ability to capture and distinguish small-target features in complex scenes. The structurally re-parameterized VGG architecture simplifies the network structure while improving detection accuracy. In addition, MGD is employed to transfer discriminative knowledge from a teacher model to a student model, thereby strengthening feature representation and generalization capability. Experiments on the officially released Road Defect Detection 2022 (RDD2022) dataset show that the proposed model improves mean average precision (mAP) by 3.4% over the baseline model. On the Street View Image Dataset for Automated Road Damage Detection (SVRDD), the model achieves a 5.9% mAP improvement. These results demonstrate that the proposed framework effectively improves road defect detection performance across both datasets and provides a feasible approach for robust detection in complex road environments.
- 5 September 2026
- Published by Elsevier in Neural Networks
- Vol. 205
Position-aware attentional neural network for review-based recommendation
- 5 September 2026
- Published by Elsevier in Neural Networks
- Vol. 205
Position-aware attentional neural network for review-based recommendation
- 22 February 2026
- Published by Springer Nature
- p. 34-42
Railway Intrusion Small Object Detection Based on Improved YOLOv8
- 22 February 2026
- Published by Springer Nature
- p. 34-42
Railway Intrusion Small Object Detection Based on Improved YOLOv8
- Open Access
- 13 February 2026
- Published by Association for Computing Machinery (ACM) in Formal Aspects of Computing
- Vol. 38 (1) , 1-48
Decidability of Liveness on the TSO Memory Model
In this article, we consider a special class of liveness properties for systems consisting of concurrent objects. These properties ensure the termination of methods calls under certain fairness assumptions and thus the progress of the execution. Liveness properties are defined for concurrent objects and they typically include lock-freedom , wait-freedom , deadlock-freedom , starvation-freedom, and obstruction-freedom . It is known that these five liveness properties are decidable for sequential consistency (SC) memory model of finite-state programs with a bounded number of processes. However, the problem of decidability of liveness for finite state concurrent programs running on relaxed memory models remains open. In this article, we address the decidability problem of liveness properties of concurrent objects for the total store order (TSO) memory model which is used in the x86 architecture. In particular, we prove that for a bounded number of processes, lock-freedom, wait-freedom, deadlock-freedom and starvation-freedom are undecidable, and that obstruction-freedom is decidable on TSO for a bounded number of processes. Further on, we investigate the verification problem of k -bounded wait-freedom , a bounded version of wait-freedom, and show that for each bound k , the problem of checking k -bounded wait-freedom is decidable on TSO for a bounded number of processes. We show that the complexity for checking obstruction-freedom and checking k -bounded wait-freedom are both non-primitive recursive. We also discover an interesting difference between liveness on TSO and that on SC. Our finding is that wait-freedom implies k -bounded wait-freedom for some k on SC memory model, but this implication does not hold on the TSO model. We prove this by generating a concrete object on TSO that is wait-free but not k -bounded wait-free for any k .
- 13 February 2026
- Published by Association for Computing Machinery (ACM) in Formal Aspects of Computing
- Vol. 38 (1) , 1-48
Open AccessDecidability of Liveness on the TSO Memory Model
In this article, we consider a special class of liveness properties for systems consisting of concurrent objects. These properties ensure the termination of methods calls under certain fairness assumptions and thus the progress of the execution. Liveness properties are defined for concurrent objects and they typically include lock-freedom , wait-freedom , deadlock-freedom , starvation-freedom, and obstruction-freedom . It is known that these five liveness properties are decidable for sequential consistency (SC) memory model of finite-state programs with a bounded number of processes. However, the problem of decidability of liveness for finite state concurrent programs running on relaxed memory models remains open. In this article, we address the decidability problem of liveness properties of concurrent objects for the total store order (TSO) memory model which is used in the x86 architecture. In particular, we prove that for a bounded number of processes, lock-freedom, wait-freedom, deadlock-freedom and starvation-freedom are undecidable, and that obstruction-freedom is decidable on TSO for a bounded number of processes. Further on, we investigate the verification problem of k -bounded wait-freedom , a bounded version of wait-freedom, and show that for each bound k , the problem of checking k -bounded wait-freedom is decidable on TSO for a bounded number of processes. We show that the complexity for checking obstruction-freedom and checking k -bounded wait-freedom are both non-primitive recursive. We also discover an interesting difference between liveness on TSO and that on SC. Our finding is that wait-freedom implies k -bounded wait-freedom for some k on SC memory model, but this implication does not hold on the TSO model. We prove this by generating a concrete object on TSO that is wait-free but not k -bounded wait-free for any k .
- 9 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-14
Enhanced Target Detection: Fusion of SPD and CoTC3 Within YOLOv5 Framework
High definition remote sensing image recognition is of great significance and can be widely used in urban planning, land use and other fields. Existing models focus excessively on texture and fine-grained features of the image during training, ignoring broader contextual information in the feature map, which reduces the ability of the model to recognise the target. In order to improve the detection accuracy and robustness, we introduce an enhanced version of the YOLOv5 framework specifically tailored for analyzing remote sensing imagery characterized by diminutive targets and reduced resolution. The model’s perceptual capacity is augmented through the inclusion of the Space-to-depth (SPD) element, alongside the adoption of null and depth-wise separable convolutions to better capture target attributes. Furthermore, we introduce the Contextual Transformer-Concentrated Comprehensive Convolution (CoTC3) module, seamlessly integrated into the YOLOv5’s core architecture. This innovative module enables the model to capitalize on rich contextual information between neighbouring keys, leading to more optimal feature representations and subsequently enhancing detection accuracy. The model undergoes refinements, particularly in the loss function, to bolster its efficacy further. Evaluated across three distinct datasets encompassing 34 categories of targets and a collective count of 382,221 objects, the revised model exhibits notable enhancements in accuracy and robustness in target detection. Notably, the accuracy in identifying vehicles and bridges improves by 10.1% and 11.3% respectively, with the overall accuracy rates ascending to 93.5% (an increment of 2.3%), 88.1%(an increment of 2.9%) and 71.2% (an increment of 6.3%) in comparison to the baseline model.
- 9 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-14
Enhanced Target Detection: Fusion of SPD and CoTC3 Within YOLOv5 Framework
High definition remote sensing image recognition is of great significance and can be widely used in urban planning, land use and other fields. Existing models focus excessively on texture and fine-grained features of the image during training, ignoring broader contextual information in the feature map, which reduces the ability of the model to recognise the target. In order to improve the detection accuracy and robustness, we introduce an enhanced version of the YOLOv5 framework specifically tailored for analyzing remote sensing imagery characterized by diminutive targets and reduced resolution. The model’s perceptual capacity is augmented through the inclusion of the Space-to-depth (SPD) element, alongside the adoption of null and depth-wise separable convolutions to better capture target attributes. Furthermore, we introduce the Contextual Transformer-Concentrated Comprehensive Convolution (CoTC3) module, seamlessly integrated into the YOLOv5’s core architecture. This innovative module enables the model to capitalize on rich contextual information between neighbouring keys, leading to more optimal feature representations and subsequently enhancing detection accuracy. The model undergoes refinements, particularly in the loss function, to bolster its efficacy further. Evaluated across three distinct datasets encompassing 34 categories of targets and a collective count of 382,221 objects, the revised model exhibits notable enhancements in accuracy and robustness in target detection. Notably, the accuracy in identifying vehicles and bridges improves by 10.1% and 11.3% respectively, with the overall accuracy rates ascending to 93.5% (an increment of 2.3%), 88.1%(an increment of 2.9%) and 71.2% (an increment of 6.3%) in comparison to the baseline model.
- 16 October 2023
- Published by Elsevier in Expert Systems with Applications
- Vol. 238
Mineral identification based on natural feature-oriented image processing and multi-label image classification
- 16 October 2023
- Published by Elsevier in Expert Systems with Applications
- Vol. 238
Mineral identification based on natural feature-oriented image processing and multi-label image classification
- Open Access
- 29 August 2022
- Published by MDPI AG in Journal of Sensor and Actuator Networks
- Vol. 11 (3) , 50
A Review of Artificial Intelligence Technologies in Mineral Identification: Classification and Visualization
Artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine capable of responding in a manner similar to human intelligence. Research in this area includes robotics, language recognition, image identification, natural language processing, and expert systems. In recent years, the availability of large datasets, the development of effective algorithms, and access to powerful computers have led to unprecedented success in artificial intelligence. This powerful tool has been used in numerous scientific and engineering fields including mineral identification. This paper summarizes the methods and techniques of artificial intelligence applied to intelligent mineral identification based on research, classifying the methods and techniques as artificial neural networks, machine learning, and deep learning. On this basis, visualization analysis is conducted for mineral identification of artificial intelligence from field development paths, research hot spots, and keywords detection, respectively. In the end, based on trend analysis and keyword analysis, we propose possible future research directions for intelligent mineral identification.
- 29 August 2022
- Published by MDPI AG in Journal of Sensor and Actuator Networks
- Vol. 11 (3) , 50
Open AccessA Review of Artificial Intelligence Technologies in Mineral Identification: Classification and Visualization
Artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine capable of responding in a manner similar to human intelligence. Research in this area includes robotics, language recognition, image identification, natural language processing, and expert systems. In recent years, the availability of large datasets, the development of effective algorithms, and access to powerful computers have led to unprecedented success in artificial intelligence. This powerful tool has been used in numerous scientific and engineering fields including mineral identification. This paper summarizes the methods and techniques of artificial intelligence applied to intelligent mineral identification based on research, classifying the methods and techniques as artificial neural networks, machine learning, and deep learning. On this basis, visualization analysis is conducted for mineral identification of artificial intelligence from field development paths, research hot spots, and keywords detection, respectively. In the end, based on trend analysis and keyword analysis, we propose possible future research directions for intelligent mineral identification.
- 22 July 2022
- Published by Elsevier in Computers & Security
- Vol. 121
A survey on adversarial attacks in computer vision: Taxonomy, visualization and future directions
- 22 July 2022
- Published by Elsevier in Computers & Security
- Vol. 121
A survey on adversarial attacks in computer vision: Taxonomy, visualization and future directions