Yi Ding

Dr. Yi Ding


Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

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

Representative Publications

View All Publications
7 October 2023 in Applied Sciences
Verified
Journal Article
Open access
PDF file
XXXinjian Xiang
Xinjian Xiang
HHHaibin Hu
Haibin Hu
YDYi Ding
Yi Ding
YZYongping Zheng
Yongping Zheng
善吴善宝 吴
善宝 吴
This study proposes a GC-YOLOv5s crack-detection network of UAVs to work out several issues, such as the low efficiency, low detection accuracy caused by shadows, occlusions and low contrast, and influences due to road noise in the classic crack-detection methods in the complicated traffic routes. A Focal-GIOU loss function with a focal loss has been introduced in this proposed algorithm, which is applied to address the issue of the imbalance of difficult and easy samples in crack images. Meanwhile, the original localization loss function CIOU is replaced by a GIOU loss function that is more suitable for irregular target (crack) detection. In order to improve the ability of the modified model of representing the features, a Transposed Convolution layer is simultaneously added in place of the original model’s upsampling layer. According to the advantage of computing resources of the Ghost module, the C3Ghost module is applied to decrease the amount of network parameters while maintaining adequate feature representation. Additionally, a lightweight module, CSPCM, is designed with the Conmix module and the Ghost concept, which successfully reduces the model parameters and zooms out the volume. At the same time, this modified module can have enough detection accuracy, and it can satisfy the requirements of UAV detection of small models and rapidity. In order to prove the model’s performance, this study has established a new UAV road-crack-detection dataset (named the UMSC), and has conducted extensive trials. To sum up, the precision of GC-YOLOv5s has increased by 8.2%, 2.8%, and 3.1%, respectively, and has reduced the model parameters by 16.2% in comparison to YOLOv5s. Furthermore, it outperforms previous YOLO comparison models in Precision, Recall, mAP_0.5, mAP_0.5:0.95, and Params.

25 May 2024 in Scientific Reports
Verified
Journal Article
Open access
PDF file
YDYi Ding
Yi Ding
KWKai Wang
Kai Wang
XWXiaojun Wu
Xiaojun Wu
XXXinjian Xiang
Xinjian Xiang
XTXianqiang Tang
Xianqiang Tang
YZYong Zhang
Yong Zhang
To address several common problems of finger vein recognition, a lightweight finger vein recognition algorithm by means of a small sample has been proposed in this study. First of all, a Gabor filter is applied to deal with the images for the purpose of that these processed images can simulate a kind of situation of finger vein at low temperature, such that the generalization ability of the algorithm model can be improved as well. By cutting down the amount of convolutional layers and fully connected layers in VGG-19, a lightweight network can be given. Meanwhile, the activation function of some convolutional layers is replaced to protect the network weight that can be updated successfully. After then, a multi-attention mechanism is introduced to the modified network architecture to result in improving the ability of extracting important features. Finally, a strategy based on transfer learning has been used to reduce the training time in the model training phase. Honestly, it is obvious that the proposed finger vein recognition algorithm has a good performance in recognition accuracy, robustness and speed. The experimental results show that the recognition accuracy can arrive at about 98.45%, which has had better performance in comparison with some existing algorithms.

4 November 2021 in Symmetry
Verified
Journal Article
Open access
PDF file
YZYong Zhang
Yong Zhang
HWHao Wang
Hao Wang
YDYi Ding
Yi Ding
BHBeiping Hou
Beiping Hou
In this paper, we propose a CPG (central pattern generator) network control system using motor dynamics for the gait planning of a quadruped robot with a trot walking pattern to climb up and down a slope and turn back and follow the symmetry of route. The CPG unit model, which includes two DC motors model, has the ability to generate the periodic joint angle with complex-value parameters. Through plural feedback parameters, the CPG network can adjust the frequency and amplitude of an internal neuron model such as a robot meeting an irregular surface of a road. Using the stride length and frequency of robot joint angles, the distance of walking with a trot pattern can be calculated. In order to confirm the validity of the proposed control system, a quadruped robot is produced to implement the adaptive walking system.

22 July 2024 in IEEE Access
Verified
Journal Article
Open access
YDYi Ding
Yi Ding
PTPusheng Tang
Pusheng Tang
JLJinrong Li
Jinrong Li
CZCheng Zeng
Cheng Zeng
YSYongzhi Sun
Yongzhi Sun
WLWei Liu
Wei Liu
Movement disorder patients are usually unable to easily communicate with the outside world. Therefore, a novel approach for EOG character input based on DC-CNN is proposed in this paper. Users only need to control the eyeball movement and eye blinking so as to manipulate the interactive interface on the screen for character input. The signal acquisition device used in this study adopts a dual-channel acquisition method. The EOG signals are preprocessed, converted to digital signals, and filtered for noise using bandpass filtering and wavelet transform. The filtered and denoised signals are input into the proposed novel approach for training the classification model. During the real-time interaction, the EOG signals are input into the trained model for classification. Next, the classification result is used as the input signal for a virtual keyboard which is compiled by PYQT5. After that, the virtual keyboard can generate corresponding responses depending on different input signals to achieve the character input. The model achieved a high classification accuracy of 99.3% during training and 98.5% in real-time classification, significantly outperforming existing methods. Furthermore, the accuracy of single character input and paragraph input can arrive at 98.2% and 97.6% as well, respectively. This novel DC-CNN approach offers a reliable and user-friendly communication method for movement disorder patients, enhancing their ability to interact with the world.

Verified
Journal Article
Open access
PDF file
ZMZhongliang Meng
Zhongliang Meng
YCYun Chen
Yun Chen
YLYanjun Liu
Yanjun Liu
YDYi Ding
Yi Ding
The mooring system not only plays a vital role in keeping wave energy generators floating stably, but also affects the success of engineering design. Combining wave force theory and the hydrological data obtained from the field measurements of a certain sea area in the Bohai Sea, the Stokes second-order wave theory was adopted to design the mooring system of a new type of power-generating device. At the same time, the study uses the Aqwa software to gather the dynamic data of a power-generating device in a real test, and then makes models and carries out regular wave tests so as to verify the viability of the mooring system and the stability of the whole power-generating device. All of this work will provide a theoretical basis for the manufacture of an engineering prototype and its reliable supply of power.

21 August 2024 in IEEE Access
Verified
Journal Article
Open access
XXXinjian Xiang
Xinjian Xiang
XCXizhao Chen
Xizhao Chen
XTXiaohui Tang
Xiaohui Tang
QLQiuxia Luo
Qiuxia Luo
YDYi Ding
Yi Ding
Speed bumps, as a crucial road safety infrastructure, if they have defects, can directly impact the accident rates on specific road sections. Therefore, this study proposes a method for detecting speed bump multiclass defects using an improved YOLOv5s model. This model modifies the pooling method in the backbone layer to Spatial Pyramid Pooling Fast (SimSPPF) to enhance the detection speed while improving the capability of feature extraction from images. In the neck layer, firstly, the upsampling method was changed to Convolutional Transpose (ConvTranspose) to increase the accuracy of small target detection. Then, the Contextual operation network (CotNet) module was integrated with the original C3 module, enhancing the model’s ability to recognize different types of defects by obtaining better global features. Finally, the convolution module was restructured to a Recursive Gated Convolution (gnconv) module, designed to maximize the model’s capacity to capture complex multi-scale, multiclass image features. Additionally, a new data augmentation method (CR) was proposed for data enhancement and balancing of the samples. Experimental results show that the improved YOLOv5s algorithm achieved an accuracy of 97.7%, a recall rate of 91.9%, and a mean precision of 96.4%, while maintaining a parameter size of only 8.8M. Compared to other YOLO detection models, the improved model exhibits high accuracy, high confidence, and a low parameter count.