Abstract
This paper investigates a critical challenge of detecting prohibited items in student dormitories, essential for maintaining safety, health, and compliance. Traditional manual inspections are inefficient, intrusive, and prone to human errors. To overcome these, we propose DormGuardNet, a novel deep learning model specifically designed for detecting prohibited items in student dormitory environments. DormGuardNet leverages ghost convolution to boost performance while reducing computational complexity. The model was implemented within the You Only Look Once (YOLO) framework, ensuring fast and highly accurate object detection. To address the lack of an existing dataset for this task, we developed PISD (Prohibited Items in Student Dormitories). Extensive experiments show that DormGuardNet achieves competitive results compared to several versions of YOLO on various evaluation metrics. Both quantitative and qualitative results presented in this paper highlight the efficiency and effectiveness of our proposed approach. Source code and dataset are available on github.com/muzahidai/DormGuardNet/.