Student Classroom Behavior Detection Dataset Overview

Dataset Name

CampusGuard

Dataset Description

CampusGuard is a dataset specifically annotated and categorized for student behaviors in campus environments, intended to provide abundant training samples for improving the YOLOv8 model. It contains five primary categories: “Mobile‑Phone” (using a phone), “No‑Helmet” (without a helmet), “Sleeping”, “Triples” (group interactions of three students), and “Violence”. These categories encompass typical in‑class and out‑of‑class behaviors and reflect the multifaceted aspects of campus safety and student conduct management.

Dataset Construction

• Diversity & Realism: The construction process emphasizes both diversity and realism, ensuring the samples faithfully represent various behaviors observed on campuses.
• Meticulous Annotation: Each category is carefully annotated to provide clear and accurate training data for models.
• Source Diversity: Samples are gathered from multiple campuses, covering different times and environments, enhancing representativeness and model adaptability.

Dataset Categories

• Mobile‑Phone: Various scenarios of students using mobile phones during class.
• No‑Helmet: Students riding bicycles on campus without helmets.
• Sleeping: Instances of students dozing off in class.
• Triples: Interactions among three students, reflecting cooperation or discussion.
• Violence: Monitoring of violent incidents on campus, aiming for early detection to ensure safety and mental health.

Dataset Application

By detecting and analyzing these behaviors, instructors can promptly assess student attention, classroom discipline, and potential safety hazards, thereby taking appropriate interventions. The dataset’s breadth and richness provide a solid foundation for model training, effectively enhancing model generalization and practical utility.

Dataset Statistics

• Images: 3,345
• Categories: 5
• Category Names: [Mobile‑Phone, No‑Helmet, Sleeping, Triples, Violence]

Significance

Research on student classroom behavior detection based on an improved YOLOv8 model holds both theoretical importance and wide practical applications. By exploring and applying this technology, new ideas and methods can be offered for educational management, promoting technological innovation and development in the education sector.