EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Classroom Behavior Detection and Object Tracking Technology for College Students Based on YOLOv7‐Tiny Model

Lili Guan, Bo Fu

Modelling and Simulation in Engineering · 2026-01 · 2026 (1)

DOI: 10.1155/mse/3200276 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Modelling and Simulation in Engineering, online erschienen am 13.08.2026. Verfasst von Lili Guan und 1 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

As the core scene of knowledge transmission, the demand for dynamic monitoring in university classrooms is increasing day by day. It needs to solve the problems of insufficient detection accuracy, trajectory breakage, and false matching caused by dense student distribution, diverse behavior postures, and multiple occlusions. To raise the accuracy and real‐time performance of classroom behavior detection and object tracking, a detection method based on an improved You Only Look Once Version 7‐tiny (YOLOv7‐tiny) model is developed. The method uses deformable convolution to enhance the adaptability of object deformation and combines ghost convolution compression to calculate redundancy. Coordinate attention mechanism is introduced to enhance target localization during the tracking phase, while designing a scalable intersection and union matching strategy that integrates multidimensional loss. The experiment is based on the MS COCO and MOT17 datasets, and compares the performance of mainstream detection models and tracking models. The findings denote that the raised classroom behavior detection model has an average accuracy of 91.2% in 10 testing scenarios, which is 2.1% higher than the current advanced detection methods. The average frame rate of the test is 393 FPS. In the tracking task, the average accuracy of multitarget tracking is 76.5%, the amount of identity document (ID) switches is reduced by 74.7% compared with the baseline, and it is 74.5% in high‐density occlusion scenes, which verifies the effectiveness of multidimensional constraint matching. Through dynamic feature modeling and lightweight architecture optimization, high‐precision and low latency classroom behavior detection and target tracking have been achieved, providing an efficient and reliable technical solution for educational intelligence.

Zitieren

APA

Guan, L., & Fu, B. (2026). Classroom Behavior Detection and Object Tracking Technology for College Students Based on YOLOv7‐Tiny Model. Modelling and Simulation in Engineering, 2026(1). https://doi.org/10.1155/mse/3200276

Vancouver

Guan L, Fu B. Classroom Behavior Detection and Object Tracking Technology for College Students Based on YOLOv7‐Tiny Model. Modelling and Simulation in Engineering. 2026;2026(1). doi:10.1155/mse/3200276

BibTeX
@article{Guan2026,
  author = {Guan, Lili and Fu, Bo},
  title = {Classroom Behavior Detection and Object Tracking Technology for College Students Based on YOLOv7‐Tiny Model},
  journal = {Modelling and Simulation in Engineering},
  year = {2026},
  volume = {2026},
  number = {1},
  doi = {10.1155/mse/3200276}
}
Kostenlos starten →