Xianrang Shi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xianrang Shi has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Hengyu Zhang
Nanjing University of Aeronautics and Astronautics, Nanjing, China
2 shared publications - Rong Wang
Nanjing University of Aeronautics and Astronautics, Nanjing, China
2 shared publications Tinglun Song
Chery Automobile CO., Ltd, Wuhu, P. R. China
2 shared publications- Tinglun Song
Chery Automobile Co., Ltd, WuHu, China
2 shared publications Yan Ti
College of Energy and Power, Nanjing University of Aeronautics and Astronautics, Nanjing, P. R. China
2 shared publications- Yan Ti
School of Automotive and Traffic Engineering, Jiangsu University of Technology, Changzhou, China
2 shared publications
Latest Publications
- 28 October 2024
- Published by Springer Nature in Machine Vision and Applications
- Vol. 35 (6) , 1-19
Slope-embedded ViT-based model for lane line detection under occlusions
- 28 October 2024
- Published by Springer Nature in Machine Vision and Applications
- Vol. 35 (6) , 1-19
Slope-embedded ViT-based model for lane line detection under occlusions
- 20 September 2023
- Published by World Scientific Pub Co Pte Ltd in International Journal of Pattern Recognition and Artificial Intelligence
- Vol. 37 (12)
LHFNet: A Fast and Accurate Object Detector Based on CenterNet
This paper proposes LHFNet, an improved object detection method based on CenterNet, which achieves a better speed-accuracy trade-off. The main contributions of this paper are as follows: We design a lightweight Hourglass network to reduce the number of parameters and computations of CenterNet, thus improving the detection speed. We introduce an intermediate feature map fusion module to enhance the feature extraction capability of Hourglass, which compensates for the possible performance degradation caused by the network simplification. We apply an attention module to the final heat map to improve the image feature representation. We design and optimize a series of loss functions to train LHFNet effectively. Through the research in this paper, we demonstrate that the feature map fusion and the attention module greatly improve the performance of LHFNet. We evaluate our method on the COCO dataset and achieve a high accuracy at a fast speed (Tesla V100), which outperforms other object detection methods in terms of both accuracy and efficiency.
- 20 September 2023
- Published by World Scientific Pub Co Pte Ltd in International Journal of Pattern Recognition and Artificial Intelligence
- Vol. 37 (12)
LHFNet: A Fast and Accurate Object Detector Based on CenterNet
This paper proposes LHFNet, an improved object detection method based on CenterNet, which achieves a better speed-accuracy trade-off. The main contributions of this paper are as follows: We design a lightweight Hourglass network to reduce the number of parameters and computations of CenterNet, thus improving the detection speed. We introduce an intermediate feature map fusion module to enhance the feature extraction capability of Hourglass, which compensates for the possible performance degradation caused by the network simplification. We apply an attention module to the final heat map to improve the image feature representation. We design and optimize a series of loss functions to train LHFNet effectively. Through the research in this paper, we demonstrate that the feature map fusion and the attention module greatly improve the performance of LHFNet. We evaluate our method on the COCO dataset and achieve a high accuracy at a fast speed (Tesla V100), which outperforms other object detection methods in terms of both accuracy and efficiency.
- Open Access
- 21 July 2023
- Published by Springer Nature in Research Square
ViT with diamond patches for lane line detection under occlusion
Lane line detection based on deep learning has achieved good results in common scenarios. However, it is challenging to detect lane lines in extreme occlusion scenes where the visual clues are severely missing. To address this problem, we propose a novel method that leverages Vision Transformer (ViT) for de-occlusion and feature fusion. Specifically, we first design a ViT-based model to reconstruct the occluded lane lines from the input image. Then, we extract the feature map of the model and fuse it with the original image feature map. Finally, we use the fused feature map to detect the lane lines in a robust manner. Moreover, we introduce a sensitivity loss function that measures the error of each pixel and considers the coordinate difference between pixels. Experiments show that our sensitivity loss function can improve the performance of lane line detection. We evaluate our method on three benchmark datasets: TuSimple, CULane and CurveLanes. The results demonstrate that our method outperforms the existing methods in terms of accuracy and F1-score on all these datasets.
- 21 July 2023
- Published by Springer Nature in Research Square
Open AccessViT with diamond patches for lane line detection under occlusion
Lane line detection based on deep learning has achieved good results in common scenarios. However, it is challenging to detect lane lines in extreme occlusion scenes where the visual clues are severely missing. To address this problem, we propose a novel method that leverages Vision Transformer (ViT) for de-occlusion and feature fusion. Specifically, we first design a ViT-based model to reconstruct the occluded lane lines from the input image. Then, we extract the feature map of the model and fuse it with the original image feature map. Finally, we use the fused feature map to detect the lane lines in a robust manner. Moreover, we introduce a sensitivity loss function that measures the error of each pixel and considers the coordinate difference between pixels. Experiments show that our sensitivity loss function can improve the performance of lane line detection. We evaluate our method on three benchmark datasets: TuSimple, CULane and CurveLanes. The results demonstrate that our method outperforms the existing methods in terms of accuracy and F1-score on all these datasets.
- 1 October 2022
- Published by World Scientific Pub Co Pte Ltd in International Journal of Pattern Recognition and Artificial Intelligence
- Vol. 36 (13)
CenterNet Plus: Multiscale Prediction with FPN for CenterNet
Although the CenterNet object detection method can be used to achieve high accuracy, it cannot be used to detect objects with overlapping or almost overlapping centers. However, a good trade-off between detection speed and accuracy cannot be achieved when utilizing this method. We improved a series of backbones using CenterNet, output the fusion of multiple feature maps in backbone, and used the feature pyramid network (FPN) mechanism for multiscale object detection. Additionally, we improved the head of the detector and considered adding intersection over union (IoU) branches. Finally, the loss function was improved to improve detection accuracy. Based on the above research, the CenterNet Plus object detection method is proposed in this paper. Through experiments on the COCO dataset, it can be seen that the use of a multiscale FPN mechanism not only solves the problem of center point overlap but also helps improve the accuracy of detection. CenterNet Plus can be used to greatly improve the detection accuracy on the premise of having a higher detection speed.
- 1 October 2022
- Published by World Scientific Pub Co Pte Ltd in International Journal of Pattern Recognition and Artificial Intelligence
- Vol. 36 (13)
CenterNet Plus: Multiscale Prediction with FPN for CenterNet
Although the CenterNet object detection method can be used to achieve high accuracy, it cannot be used to detect objects with overlapping or almost overlapping centers. However, a good trade-off between detection speed and accuracy cannot be achieved when utilizing this method. We improved a series of backbones using CenterNet, output the fusion of multiple feature maps in backbone, and used the feature pyramid network (FPN) mechanism for multiscale object detection. Additionally, we improved the head of the detector and considered adding intersection over union (IoU) branches. Finally, the loss function was improved to improve detection accuracy. Based on the above research, the CenterNet Plus object detection method is proposed in this paper. Through experiments on the COCO dataset, it can be seen that the use of a multiscale FPN mechanism not only solves the problem of center point overlap but also helps improve the accuracy of detection. CenterNet Plus can be used to greatly improve the detection accuracy on the premise of having a higher detection speed.