A System for Real-Time Detection of Abandoned Luggage

In public spaces such as airports, one of the primary security risks is abandoned luggage, which can potentially threaten public safety and disrupt normal business operations. This is a challenging task because these spaces are often crowded, with many people moving in different directions, carrying or pulling luggage, or standing in lines or waiting with luggage.In this paper we propose a system for automatic detection of abandoned luggage in an airport recorded by surveillance cameras, in real time, using a customized YOLOv11-l model and a proposed algorithm for detecting unattended luggage. The system uses the OpenCV library for video processing of the recorded material, a detector, and an algorithm that analyzes the movement of the person and the luggage and evaluates their spatial and temporal relationships to determine whether the luggage is truly abandoned.We used different popular architectures of deep convolutional neural networks for object detection such as Yolov8 and Yolov11 and DETR encoder-decoder transformer with a ResNet-50 deep convolutional backbone, and we fine-tuned them on our custom dataset and compared their performance in detecting people and luggage in surveillance scenes recorded by the airport surveillance camera. Detection of people and luggage recorded by the airport surveillance camera are significantly improved by the fine-tuned model on our custom dataset. Both YOLOv8 and YOLOv11 fine-tuned models achieved excellent results on a demanding set consisting of only small and medium-sized objects in real time, achieving precision of over 70% from 3% mAP, while their precision for medium-sized objects was over 93%. However, the YOLOv11-l model achieves the highest precision in detecting small objects of 69%, which is why we selected it as a component of the abandoned baggage detection system. The abandoned baggage detection algorithm was tested in various scenarios in which baggage can be left and in situations that can be potentially suspicious and shows promising results.