Drone-Based Real-Time Crowd Density Estimation Using Deep Learning Techniques

Abstract

Crowd management is really important for preventing accidents and keeping people safe during events like religious gatherings, festivals and political rallies.Theoldwaysofwatchingpeoplelikewithsecurity cameras are not good enough because they do not cover everythingtheyarenotflexible.Theydonotrespondfast enough. This project is about a system that uses drones and Artificial Intelligence to watch crowds and figure out if there are many people in one place especially in areas that are open or might be dangerous. A drone with a camera takes pictures of the crowd from the air and then acomputerprogramlikeYOLOv8looksatthesepictures to find and count people calculate how crowded it is and see how people are moving.The system always checks this information against safety rules to see if there are many people. If it finds out that there are many people it sends out warnings and updates a special board that the people in charge can look at to take action right away. This system is better than the ways because it can see more and it is more flexible. It also means that we donot need many people on the ground to watch the crowd, which helps us understand what is happening better and reduces the risk of people panicking or running intoeach other. Crowd management is really important. This system can help with that. The drone system can help keep people safe during events, like festivals and rallies.

Introduction

The paper proposes a real-time crowd monitoring system using drones, CCTV cameras, and deep learning techniques to improve safety during large public gatherings such as festivals, protests, and emergencies. Traditional surveillance methods rely heavily on manual monitoring and fixed cameras, which are often slow, inaccurate, and unable to respond quickly to dynamic crowd situations.

The system uses drone-based aerial surveillance combined with computer vision models to detect and analyze crowd behavior in real time. Video data is processed by extracting frames, preprocessing them, and then applying the YOLOv8 model for detecting individuals. After detection, DeepSORT tracking assigns unique IDs to people to track movement and avoid double counting. This enables accurate estimation of crowd size, density, and flow patterns.

The system also includes crowd analysis and anomaly detection, identifying risky situations such as sudden crowd surges, abnormal movement, running, or falling. When such events are detected, the system triggers real-time alerts through a dashboard so authorities can respond quickly and prevent incidents like stampedes.

The architecture includes stages like data acquisition (drones and CCTV), preprocessing, detection, tracking, analysis, alert generation, and result storage. A dashboard provides visual insights such as crowd density, movement trends, and warnings.

However, the system faces challenges such as difficulty in tracking dense crowds, drone limitations (battery life and stability), weather interference, high computational requirements, privacy concerns, and legal restrictions. Despite these issues, it offers a scalable and efficient solution for modern crowd management.

From a security and privacy perspective, the system emphasizes encrypted data transmission, access control, secure storage, and minimal data retention. It also suggests privacy-preserving techniques like face blurring and strict compliance with drone surveillance regulations.

Conclusion

This project is about using drones to watch crowds in time. It is an efficient way to dothis. The system uses models to find objects and track them. This means it can tell who is there how many people are in the crowd and what they are doing. Drones can see a lot of people at once which’s hard to do with regular cameras. The system can also find problems like many peopleinoneplaceorpeoplemovingquickly. This helps the police know what is happening so they can stop accidents like people getting hurt in a crowd. The system is automatic so peopledonothavetowatcheverythingall thetime.Thismeanstherearemistakesand it works better. The project is a way to manage crowds because it is big enough to work for manypeopleitdoesnotcosttoomuchandit is smart. It is useful for things, like concerts,big gatherings and emergencies. The crowd monitoring system using drones is an idea because it helps keep people safe. The drones and the system work together to make sure everything is okay. The project is an exampleofhowcrowdmonitoringcanbedoneinasmart way.

References

This project is about using drones to watch crowds in time. It is an efficient way to dothis. The system uses models to find objects and track them. This means it can tell who is there how many people are in the crowd and what they are doing. Drones can see a lot of people at once which’s hard to do with regular cameras. The system can also find problems like many peopleinoneplaceorpeoplemovingquickly. This helps the police know what is happening so they can stop accidents like people getting hurt in a crowd. The system is automatic so peopledonothavetowatcheverythingall thetime.Thismeanstherearemistakesand it works better. The project is a way to manage crowds because it is big enough to work for manypeopleitdoesnotcosttoomuchandit is smart. It is useful for things, like concerts,big gatherings and emergencies. The crowd monitoring system using drones is an idea because it helps keep people safe. The drones and the system work together to make sure everything is okay. The project is an exampleofhowcrowdmonitoringcanbedoneinasmart way.

Copyright

Copyright © 2026 Prof. S. V. Jagzap, Pragati Katmore, Sanika Nimbalkar, Mayuri Chavan. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.