Li Qin Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Li Qin Liu has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Hongli Yuan
Big Data and Artificial Intelligence College, Anhui Xinhua University, Hefei, China
3 shared publications - Fu Cheng Zhang
Anhui Xinhua University, China
2 shared publications - Yan Chen
Anhui Xinhua University, China
2 shared publications - Eric Blancaflor
School of Information Technology, Mapa University, Manila, Philippines
1 shared publication - Fucheng Zhang
Anhui Xinhua College, Hefei, Anhui, China
1 shared publication - Mideth Abisado
College of Computing and Information Technologies, National University, Manila, Philippines
1 shared publication
Latest Publications
- Open Access
- 28 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 27587-27595
Research on Pedestrian Re-Identification Based on Enhanced Residual Attention Network
Pedestrian re-identification is influenced by various complex factors such as camera angles, posture changes, and object occlusion. How to effectively extract discriminative feature information remains a key challenge. To address this issue, this paper proposes an enhanced neural network based on residual attention mechanism and random erasure. Firstly, by introducing attention mechanism in the residual network, the response of key features is effectively strengthened and redundant features are suppressed, thereby improving the discriminative performance of the network. Secondly, using random erasure as a data augmentation method effectively alleviates the impact of occlusion on the model’s recognition ability, reduces the risk of overfitting, and improves the network’s generalization performance. In addition, using triplet loss for supervised training of the network promotes better clustering performance of samples in the feature space, thereby significantly improving recognition accuracy. The experimental results show that the proposed method achieves a Rank-1 accuracy of 88.55% and an average accuracy (mAP) of 79.45% on the Market-1501 dataset, which is superior to existing methods. Verified that the proposed method can better extract discriminative pedestrian features and improve the performance of pedestrian re-identification models.
- 28 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 27587-27595
Open AccessResearch on Pedestrian Re-Identification Based on Enhanced Residual Attention Network
Pedestrian re-identification is influenced by various complex factors such as camera angles, posture changes, and object occlusion. How to effectively extract discriminative feature information remains a key challenge. To address this issue, this paper proposes an enhanced neural network based on residual attention mechanism and random erasure. Firstly, by introducing attention mechanism in the residual network, the response of key features is effectively strengthened and redundant features are suppressed, thereby improving the discriminative performance of the network. Secondly, using random erasure as a data augmentation method effectively alleviates the impact of occlusion on the model’s recognition ability, reduces the risk of overfitting, and improves the network’s generalization performance. In addition, using triplet loss for supervised training of the network promotes better clustering performance of samples in the feature space, thereby significantly improving recognition accuracy. The experimental results show that the proposed method achieves a Rank-1 accuracy of 88.55% and an average accuracy (mAP) of 79.45% on the Market-1501 dataset, which is superior to existing methods. Verified that the proposed method can better extract discriminative pedestrian features and improve the performance of pedestrian re-identification models.
- Open Access
- 28 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 99-103
Research on Course Recommendation Model Based on MLP and Attention Mechanism
Course recommendation plays a significant role in the field of education. Through personalized recommendation systems, students can more quickly and effectively find course resources that match their interests and learning needs, thereby improving learning efficiency, academic performance, and personal development. The multilayer perceptron (MLP) is a commonly used feedforward neural network structure, widely applied in tasks such as classification, regression, and feature extraction, learning complex representations of data through multiple fully connected hidden layers. The attention mechanism, which simulates human attention, can assign different attention weights to different parts when processing sequential data. It is widely used in machine translation, speech recognition, image captioning, and recommendation systems, improving model performance and effectiveness. This paper combines MLP and the attention mechanism in a model that better captures students' learning interests and behavior patterns, thereby enhancing recommendation accuracy and effectiveness. The attention mechanism dynamically adjusts the recommended course list based on the student's current learning state and needs, making the recommendations more aligned with the student's actual requirements. Experimental validation has shown that this model achieves significant improvements in recommendation effectiveness on real-world datasets, demonstrating the efficacy and superiority of the MLP and attention mechanism in the course recommendation task.
- 28 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 99-103
Open AccessResearch on Course Recommendation Model Based on MLP and Attention Mechanism
Course recommendation plays a significant role in the field of education. Through personalized recommendation systems, students can more quickly and effectively find course resources that match their interests and learning needs, thereby improving learning efficiency, academic performance, and personal development. The multilayer perceptron (MLP) is a commonly used feedforward neural network structure, widely applied in tasks such as classification, regression, and feature extraction, learning complex representations of data through multiple fully connected hidden layers. The attention mechanism, which simulates human attention, can assign different attention weights to different parts when processing sequential data. It is widely used in machine translation, speech recognition, image captioning, and recommendation systems, improving model performance and effectiveness. This paper combines MLP and the attention mechanism in a model that better captures students' learning interests and behavior patterns, thereby enhancing recommendation accuracy and effectiveness. The attention mechanism dynamically adjusts the recommended course list based on the student's current learning state and needs, making the recommendations more aligned with the student's actual requirements. Experimental validation has shown that this model achieves significant improvements in recommendation effectiveness on real-world datasets, demonstrating the efficacy and superiority of the MLP and attention mechanism in the course recommendation task.
- Open Access
- 23 December 2022
- Published by Association for Computing Machinery (ACM)
- p. 118-121
The Research on Image Classification Based on Convolutional Neural Network
Convolutional neural network (CNN) performs well in image classification and segmentation, target detection and other applications. Researchers are paying more and more attention to its powerful feature learning and feature expression abilities. Based on CIFAR10 data set, this paper used CNN for image recognition. In the experiment, the training accuracy and testing accuracy of CNN can be improved by increasing the training times and the number of full connection layers of CNN. In this paper, if training 1000 epochs, and use 2 full connections layers. the training accuracy can reach 91, and the testing accuracy can reach 90.5.
- 23 December 2022
- Published by Association for Computing Machinery (ACM)
- p. 118-121
Open AccessThe Research on Image Classification Based on Convolutional Neural Network
Convolutional neural network (CNN) performs well in image classification and segmentation, target detection and other applications. Researchers are paying more and more attention to its powerful feature learning and feature expression abilities. Based on CIFAR10 data set, this paper used CNN for image recognition. In the experiment, the training accuracy and testing accuracy of CNN can be improved by increasing the training times and the number of full connection layers of CNN. In this paper, if training 1000 epochs, and use 2 full connections layers. the training accuracy can reach 91, and the testing accuracy can reach 90.5.
- Open Access
- 23 December 2022
- Published by Association for Computing Machinery (ACM)
- p. 293-299
Simulation of Emergency Evacuation Model at Ziyun Road Subway Station
Subways are popular among the public because of their high capacity, speed, punctuality, convenience and safety. Due to the dense subway stations and relatively closed space, it is difficult to evacuate and rescue passengers in case of explosion, fire, flood and other events. If the station cannot effectively organize the evacuation of people, the panic and heavy psychological factors prevailing in the crowd. These factors often make the consequences of the accident further expand, or even cause a disaster. Taking Ziyun Road Subway Station as an example, this paper conducts research on effective crowd evacuation and evacuation techniques based on the characteristics of emergency evacuation of crowds in subway stations. Reasonable and effective evacuation measures and recommendations are proposed. Prevent further escalation of accidents.
- 23 December 2022
- Published by Association for Computing Machinery (ACM)
- p. 293-299
Open AccessSimulation of Emergency Evacuation Model at Ziyun Road Subway Station
Subways are popular among the public because of their high capacity, speed, punctuality, convenience and safety. Due to the dense subway stations and relatively closed space, it is difficult to evacuate and rescue passengers in case of explosion, fire, flood and other events. If the station cannot effectively organize the evacuation of people, the panic and heavy psychological factors prevailing in the crowd. These factors often make the consequences of the accident further expand, or even cause a disaster. Taking Ziyun Road Subway Station as an example, this paper conducts research on effective crowd evacuation and evacuation techniques based on the characteristics of emergency evacuation of crowds in subway stations. Reasonable and effective evacuation measures and recommendations are proposed. Prevent further escalation of accidents.