Yimeng Zhang

Yimeng Zhang

Auto-generated profile, based on publicly available information

Yimeng Zhang

The School of Computer Science and Technology, Key Laboratory of Smart Human-Computer Interaction and Wearable Technology of Shaanxi Province, Xidian University, Xian, China

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

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

Representative Publications

View All Publications
31 March 2025 in IEEE Transactions on Computational Social Systems
Verified
Journal Article
Closed access
MDMin Dang
Min Dang
Gang LiuGang Liu
Gang Liu
XLXike Li
Xike Li
BWBo Wan
Bo Wan
YZYimeng Zhang
Yimeng Zhang
RPRong Pan
Rong Pan
Student pose information can reflect learning status, which is significant in teaching management and evaluation. However, the traditional manual behavior recognition and analysis process is complex and slow, so counting massive classroom data is a difficult task. Therefore, using computer vision technology to accurately recognize student behavior is of great significance to teaching management and evaluation. Besides, the performance of existing behavior recognition methods is limited by problems such as dense objects and occlusion in classroom scenes. To address these issues, we propose an anchor-free object detector based on center keypoints for behavior recognition in classroom scenes. Specifically, we design a multiscale convolution neural networks (CNNs) module to alleviate the scale variation of objects through multiscale receptive fields. The head network is based on the anchor-free detection head architecture and integrates the keypoint heatmaps to accurately extract the center keypoint position through the center pooling module (CPM). the CPM can suppress irrelevant background information and effectively alleviate the missed detection of dense objects. Regressing the distances from the center keypoint of the positive region to the four sides can suppress low-quality bounding boxes so that the detector can accurately locate the object. Furthermore, during the inference stage, the heatmap scores of keypoints and center keypoints are linearly combined as novel confidence to mitigate inaccurate measurement of predicted bounding boxes. The extensive experimental performance comparison on the classroom behavior (CB) and SCB-dataset3 benchmark datasets demonstrate that the proposed behavior recognition method can accurately detect objects in classroom scenes. Compared with other current state-of-the-art methods, the proposed behavior recognition method based on anchor-free object detector is able to achieve good performance.

2 June 2025 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
YZYuanze Zhang
Yuanze Zhang
YZYimeng Zhang
Yimeng Zhang
KLKexu Li
Kexu Li
JLJinpeng Luo
Jinpeng Luo
Gang LiuGang Liu
Gang Liu
RPRong Pan
Rong Pan
As the pet industry develops, fine-grained breed recognition and individual recognition have emerged as essential components in biometric measurement systems for intelligent pet monitoring, aiming to identify the specific breed of a pet in an image and to recognize the same individual across multiple images. These capabilities lay the foundation for downstream tasks such as posture estimation and emotion analysis, supporting a wide range of real-world applications. Despite the substantial advancements achieved in existing research, two critical issues remain to be solved: the diversity of object poses affects representation in complex scenarios, and the conflict between model complexity and performance hinders application in resource-constrained conditions. To address the above issues, we propose an integrated Face-and-Body Information for Lightweight Breed and Individual Recognition (iFBI) scheme that integrates multiple pose information by a lightweight model. Specifically, a Face Alignment (FA) module and a Body Posture-Guided (BPG) module are proposed to separate face and body information from the input images, fully capturing the posture details while suppressing background areas. To further maximize the discrimination between individual samples, we design a Multi-level Representation Recognition (MRR) module that dynamically integrates complementary semantic features of face and body, consequently generating more discriminative features. In addition, a Dynamic Convolutional Model Compression (DCMC) method is implemented with an improved dual-branch backbone that significantly reduces computational costs while enhancing model performance. Extensive experiments on two self-built datasets, Pet with Fine-grained Breed Dataset (Pet-FB) and Pet with Diverse Posture Dataset (Pet-DP), and four public datasets indicate that iFBI yields superior performance in both fine-grained breed recognition and individual recognition tasks. The source code and self-built datasets—Pet-FB, Pet-DP—are available at our github repository.

30 July 2026 in Pattern Recognition
Verified
Journal Article
Closed access
YZYimeng Zhang
Yimeng Zhang
RPRong Pan
Rong Pan
YZYuanze Zhang
Yuanze Zhang
MDMin Dang
Min Dang
Gang LiuGang Liu
Gang Liu