Related Experiment Video
Updated: Mar 20, 2026
03:14
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
1.3K
SmartSport: crowd counting meets large language models for smart facility management
1Xuchang University, No. 88 Bayi Road, Weidu District, Xuchang, 461000, Henan, China.
Scientific Reports
|March 19, 2026
Summary
SmartSport uses computer vision and large language models to intelligently manage public sports facilities. This framework improves data analysis for better facility allocation and usage insights.
Area of Science:
- Computer Science
- Sports Management
- Artificial Intelligence
Background:
- Urban public sports facility management faces challenges with traditional methods like manual counting and surveys.
- Current approaches lack efficiency, real-time data, and actionable insights for intelligent decision-making.
Purpose of the Study:
- To introduce SmartSport, an intelligent management framework for public sports facilities.
- To address limitations in current facility management by integrating computer vision and large language models.
Main Methods:
- Developed CrowdVision module using a lightweight visual state space model for accurate crowd counting and localization.
- Integrated LLM-Advisor module employing large language models for analyzing crowd data with geographic and demographic information.
- Utilized prompt engineering for usage pattern recognition, supply-demand gap diagnosis, and optimization recommendations.
Main Results:
- CrowdVision module achieved 93.8% counting accuracy.
- LLM-Advisor module's recommendations scored 4.2/5.0 in practicality by domain experts.
- The framework generates comprehensive reports with problem attribution and actionable suggestions.
Conclusions:
- SmartSport offers an effective intelligent management solution for public sports facilities.
- The synergy of computer vision and large language models enhances facility utilization assessment and decision-making.
