Skip to content
Open access

An Intelligent Person Detection System for Public Spaces Using YOLOv11

2026 · The Journal of Artificial Intelligence and Emerging Technologies · 0 citations

Abstract

The increasing use of camera-based monitoring in public spaces requires person detection systems that can operate across changes in illumination, viewpoint, background complexity, crowd density, and object scale. This paper presents an intelligent person detection framework based on YOLOv11m for open-area surveillance. A custom dataset containing more than 6,000 annotated outdoor images was prepared to represent diverse scene conditions and person distributions. The images were used to train the YOLOv11m detector, which performs person localization through bounding boxes and confidence scores in a single-stage detection pipeline. The trained model was integrated with a Streamlit application to provide image upload, preprocessing, detection visualization, performance reporting, and optional person-count and occupancy information. Evaluation used Precision, Recall, mAP@50, Box Loss, and inference time. The reported final evaluation values were 89.41% Precision, 87.22% Recall, 88.21% mAP@50, and 0.499 Box Loss. A representative application inference reported demonstration environment. Training curves showed progressive improvement in Precision, Recall, and mAP@50 with a corresponding reduction in Box Loss. The resulting framework provides an integrated approach for person detection and performance analysis in varied outdoor public-space scenes.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.