Fen XU

Prof. Fen XU

orcid

Research Focus & Metrics

Academic Career Length

18Years (2009-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
20 November 2024 in Journal of Advanced Computational Intelligence and Intelligent Informatics
Verified
Journal Article
Open access
FXFen XU
Fen XU
EOEditorial Office
Editorial Office
PSPengfei Shi
Pengfei Shi
XZXiaoping Zhang
Xiaoping Zhang
Skeleton-based human action recognition has great potential for human behavior analysis owing to its simplicity and robustness in varying environments. This paper presents a spatial and temporal attention-enhanced graph convolution network (STAEGCN) for human action recognition. The spatial-temporal attention module in the network uses convolution embedding for positional information and adopts multi-head self-attention mechanism to extract spatial and temporal attention separately from the input series of the skeleton. The spatial and temporal attention are then concatenated into an entire attention map according to a specific ratio. The proposed spatial and temporal attention module was integrated with an adaptive graph convolution network to form the backbone of STAEGCN. Based on STAEGCN, a two-stream skeleton-based human action recognition model was trained and evaluated. The model performed better on both NTU RGB+D and Kinetics 400 than 2s-AGCN and its variants. It was proven that the strategy of decoupling spatial and temporal attention and combining them in a flexible way helps improve the performance of graph convolution networks in skeleton-based human action recognition.

21 March 2024 in Applied Artificial Intelligence
Verified
Journal Article
Open access
FXFen XU
Fen XU
JWJian Wang
Jian Wang
MGMinghuan Guo
Minghuan Guo
ZWZhifeng Wang
Zhifeng Wang

1 October 2024 in Sensors
Verified
Journal Article
Open access
PDF file
FXFen XU
Fen XU
CLChanghao Li
Changhao Li
FSFeihu Sun
Feihu Sun
The tracking pose of heliostats directly affects the stability and working efficiency of concentrated solar power (CSP) plants. Due to occlusion, over-exposure, and uneven illumination caused by mirror reflection, traditional image processing algorithms showed poor performances on the detection and segmentation of heliostats, which impede vision-based 3D measurement of tracking poses of heliostats. To tackle this issue, object detection using deep learning neural networks are exploited. An improved neural network based on YOLO-v5 framework has been designed to solve the on-line detection problem of heliostats. The model achieves a recognition accuracy of 99.7% for the test set, outperforming traditional methods significantly. Based on segmented results, the corner points of each heliostat are found out using Hough Transform and line intersection methods. The 3D poses of each heliostat are then solved out based on the image coordinates of specific feature points and the camera model. Experimental and field test results demonstrate the feasibility of this hybrid approach, which provides a low-cost solution for the monitoring and measurement of tracking poses of the heliostats in CSP.

15 March 2025 in Energies
Verified
Journal Article
Open access
PDF file
FXFen XU
Fen XU
YSYanpeng Sun
Yanpeng Sun
GGGuo Guo
Guo Guo
Predicting the solar flux density distribution formed by heliostats in a concentrated solar tower power (CSP) plant is important for the optimization and stable operation of a CSP plant. However, the high temperature and blackbody attribute of the receiver makes direct measurement of the concentrated solar irradiance distribution a difficult task. To address this issue, indirect methods have been proposed. Nevertheless, these methods are either costly or not accurate enough. This study proposes a ray tracing-assisted deep learning method for the prediction of the concentrated solar flux density distribution formed by a heliostat. Namely, a generative adversarial neural network (GAN) model using Monte Carlo ray tracing results as the input was built for the prediction of solar flux density distribution concentrated by a heliostat. Experiments showed that the predicted solar flux density distributions were highly consistent with the concentrated solar spots on the Lambertian target formed by the same heliostat. This ray tracing-assisted deep learning method can be extended to other heliostats in the CSP plant and pave the way for the prediction of the solar flux density distribution concentrated by the whole heliostat field in a CSP plant.

1 July 2015
Verified
Conference Paper
Closed access
FXFen XU
Fen XU
WZWang Zhen
Wang Zhen
This paper introduces a visual SLAM system based on Kinect sensor and ORB features. The RGB image output by the Kinect sensor is processed and Oriented FAST and Rotated BRIEF (ORB) features are extracted for landmark detection. The depth information output by Kinect is processed and repaired using nearest neighbor repairing algorithm to achieve a complete depth map. Locality-sensitive hashing (LSH) algorithm is used during feature association process to improve the mapping speed. Experiments show that the visual SLAM system based on Kinect and ORB feature is feasible and has an acceptable accuracy, with a good potential for indoor robot navigation tasks in an unknown environment.

10 December 2009 in JOURNAL OF ELECTRONIC MEASUREMENT AND INSTRUMENT
Verified
Journal Article
Closed access
FXFen XU
Fen XU
BXBaojin Xian
Baojin Xian
ZLZhengxi Li
Zhengxi Li