Make full use of the existing surveillance cameras on campus and, on top of traditional security monitoring, connect the Dragonfly AI Edge Computing Server with “single-line access” to achieve unified management and integrated linkage of the campus security system.
Using AI intelligent analysis technology, real-time detection is carried out for dangerous and abnormal activities at the school’s perimeter boundaries, main gate, finance office, archives room, campus cafeteria, student dormitories, and hazardous areas on campus. The system immediately pushes alarm information to administrators, enabling timely response measures and improving the comprehensive campus safety prevention level. It is suitable for intelligent upgrade and transformation of surveillance systems in kindergartens, primary schools, middle schools, high schools, and universities across the country. It changes traditional passive security into active early warning, builds a solid security system for schools, and provides strong evidence support for accident analysis and handling.
This innovative measure will strongly promote the improvement of school security work, further advance the construction of a safe and civilized campus, and lay a solid foundation for creating a safer and better educational environment.

At the same time, in order to prevent campus bullying, entrance and exit safety, controlled knives, accidental falls, gatherings, loitering, smoking, and other problems, intelligent transformation of the existing video surveillance system is carried out, making full use of existing surveillance video to build a comprehensive, three-dimensional safe campus security prevention and control system from multiple locations such as the school environment, personnel, vehicles, and sites, and realize comprehensive management functions such as early warning, alarm, business supervision, and remote supervision. Taking the implementation scenario of Safe Campus at a certain school as an example, by connecting to ordinary cameras already deployed by the school, key monitoring is carried out at the school gate, classrooms, playground, perimeter walls, and other scenarios. Combined with cloud AI capabilities such as face recognition and personnel matching, when dangerous situations are found, such as a student leaving school alone, a student being picked up by a stranger after school, or a stranger intruding into the school, automatic mobile phone message notifications are sent to the relevant homeroom teachers and teachers for handling, so as to avoid campus safety accidents in a timely manner.

Hardware Devices

Dragonfly AI Edge Computing Server

The Dragonfly AI Edge Computing Server is a self-developed lightweight, intelligent, edge-deployable intelligent terminal for video detection and recognition. Targeting the video and surveillance field, the system is preloaded with a powerful AI algorithm engine and can flexibly deploy various recognition algorithms from the cloud. It is positioned as an integrated software and hardware product for broad visual artificial intelligence applications. AI chip + AI framework + AI algorithm work together to provide efficient and flexible AI support, support intelligent analysis of high-channel-count video, and quickly implement AI applications such as face recognition, license plate recognition, person and vehicle control, boundary alerting, and smoke and fire alarms (implemented through cloud deployment), providing users with cost-effective intelligent solutions. It is widely used in various broad security application scenarios such as parks, campuses, construction sites, and communities.

Software System

Web-Based AI Edge Computing Server Management Platform

The Web-based AI Edge Computing Server Management Platform is a video risk monitoring and management system for managing and maintaining AI edge servers and camera video streams. It has modules such as risk monitoring, face recognition monitoring, host monitoring, AI host management, AI unit management, camera management, upgrade management, basic information management, alarm center, and system settings. The system has flexible configuration and management functions, can conveniently add, delete, adjust, and view AI edge servers and other devices, and also supports multiple security protection mechanisms to ensure the confidentiality and integrity of monitoring data.

The Web-based AI Edge Computing Server Management Platform has the following advantages compared with traditional management platforms:
1. Flexibility and scalability: The platform provides a flexible architecture and interfaces, allowing algorithm models and datasets to be easily added, deleted, upgraded, and configured, with better scalability.
2. Efficiency and accuracy: The platform can adjust algorithm parameters, select the best algorithm model, and so on, improving the efficiency and accuracy of algorithm operation.
3. Security and controllability: The platform provides comprehensive permission management, security protection mechanisms, and access control to ensure the security and controllability of algorithms and their data.
4. Full lifecycle management: The platform can manage the full lifecycle of algorithms and AI devices, including development, testing, deployment, and operation, ensuring traceability and sustainability throughout the entire process.
5. Data visualization: The platform provides data visualization functions, can clearly display data and results during algorithm operation, and helps users better understand algorithm effects and parameter tuning.

Streaming Media Server

Intranet environment: The streaming media server pulls the camera RTSP video stream for distribution and processing, and video surveillance can be viewed on the management-side software.
Public network environment: The local streaming media server pulls and forwards the camera RTSP video stream to the cloud streaming media server, where it is distributed and processed, and video surveillance can be viewed in the management-side software.
The streaming media server features high-concurrency processing capability, support for multiple protocols, and high security performance, with no limit on the number of channels, and can implement camera video surveillance. Specific advantages include the following:
1. Efficiency: The streaming media server uses newer technologies and protocols to optimize streaming media transmission efficiency, can send and receive audio and video data faster, and improves streaming media transmission speed and quality.
2. Scalability: The streaming media server can be deployed in a cluster according to actual needs, increasing the number of servers and improving system scalability to cope with larger user traffic and more complex application scenarios.
3. Ease of use: Streaming media servers usually have a friendly web management interface, which can help administrators more conveniently configure parameters, monitor status, manage users, and perform other operations.
4. Security: The streaming media server supports security mechanisms such as data encryption, identity authentication, and authorization, and can prevent network security threats such as attacks and DDoS, ensuring the secure transmission of streaming media data.
5. Stress testing: The streaming media server can perform stress testing, simulate high-concurrency access conditions, evaluate server performance, stability, and load capacity, and help administrators better optimize system performance and resource allocation.
6. Customizability: The streaming media server can provide flexible API interfaces and plug-in development frameworks, support the development and integration of custom functions, and meet users' customized needs.

Mobile App

The Android mobile app supports functions such as device monitoring, risk monitoring, risk notification, risk handling, device alarms, and alarm handling, making it convenient for users to view and manage surveillance cameras with their mobile phones at any time without being on site or using a computer. Other advantages are as follows:
1. Real-time monitoring: The app supports real-time monitoring of device status and video images, and can respond promptly to security events and abnormal situations.
2. Security early warning: The app supports intelligent risk monitoring and notification functions, can perform real-time analysis of key parameters such as devices, locations, and time, and provide early warnings of security risks.
3. Risk handling: The app provides detailed risk handling processes and tracking records, enabling users to respond quickly and accurately to various problems.
4. Multi-device management: The app supports simultaneous management of multiple monitoring devices, allowing users to conveniently and uniformly manage security monitoring systems in multiple locations.

Windows Desktop Reminder

Installed on commonly used Windows work computers, it helps users filter important risk messages. Users can directly enter the risk handling page through the desktop reminder tool. It offers high visibility, is intuitive and convenient, and improves work efficiency.

SMS Message Reminder

SMS notification reminders can notify users without network access and without delay, enabling users to learn the latest risk information in a timely manner, and the read rate is also higher than other methods.

Smoking Recognition

Through the AI smoking recognition algorithm, students' smoking behavior is monitored and investigated, potential safety hazards caused by smoking are discovered in a timely manner, corresponding measures are taken to prevent accidents, and the system will immediately issue an automatic early warning and notify relevant personnel for timely handling.

Face Recognition

Through the AI face recognition algorithm, students entering and leaving the school gate can be identified and recorded, effectively controlling the entry of external personnel into the campus to ensure the safety of the school campus environment. At the same time, students entering and leaving dormitories can be identified and recorded, so that abnormal situations can be quickly discovered and handled in a timely manner.

Fall Recognition

Every year, a large number of students are injured due to accidents, and the AI fall recognition algorithm can monitor the movement status of pedestrians on the campus ground, automatically identify possible fall events, and promptly send alarm signals to relevant personnel to improve campus safety levels.

Fighting Recognition

Every year, many campus violence incidents occur, posing a great threat to the physical and mental health of teachers and students. The AI fighting recognition algorithm can promptly discover and alarm fighting behavior on campus, prevent larger incidents, and improve campus safety levels.

Gathering Recognition

Through the AI gathering recognition algorithm, real-time monitoring of campus areas can promptly discover abnormal situations such as students gathering and crowding around, promptly discover abnormal student behavior on campus, prevent safety accidents, and protect campus student safety.

Area Intrusion Recognition

The AI area intrusion recognition algorithm can monitor the activity trajectories of personnel on campus in real time and promptly discover suspected intrusion behaviors such as strangers wandering on campus, quickly intercept them, and at the same time the early warning system will promptly send police dispatch information to campus administrators, strengthening campus safety prevention and emergency response.

Climbing Recognition

The AI climbing recognition algorithm can identify persons who are climbing and promptly issue early warning prompts to avoid illegal acts such as campus climbing. It can realize automatic monitoring without repeated manual intervention, reducing labor costs and risks caused by negligence.

Controlled Knife Recognition

The AI controlled knife recognition algorithm can monitor students carrying knives in real time and promptly alarm, so as to prevent dangerous incidents, effectively prevent campus violence incidents, and safeguard students' personal safety and the school's safety and stability.

Flame Recognition

Through the AI flame recognition algorithm, real-time monitoring of campus areas can quickly discover fire hazards and provide timely early warnings, safeguarding student safety. At the same time, it can improve the rescue efficiency of fire accidents and, for emergency situations involving sudden fire accidents, obtain information more quickly and take corresponding emergency measures rapidly.

Dense Smoke Recognition

Through the AI dense smoke recognition algorithm, real-time monitoring of campus areas is carried out. When a large amount of dense smoke is generated in the site, the system can provide timely early warning and issue alarm signals, allowing the school to take corresponding emergency measures and safeguard the safety of teachers and students.

Vehicle License Plate Recognition

The AI vehicle license plate recognition algorithm can quickly and efficiently identify vehicles entering and leaving the campus, including license plate number, vehicle model, color, and other information. It can automatically identify and record vehicle conditions without repeated manual intervention, reducing labor costs and risks caused by negligence.

To view more AI algorithms for safe campus scenarios, please visit: AiPlusMall AI Vision Mall - Safe Campus

Q1: There are 11 recognition algorithms. Do we need to buy all of them or can we choose?

A: You can choose. The algorithms all run on the edge server and are enabled by scenario. You can subscribe to whichever ones you use. Use face recognition and license plate recognition at the school gate, smoking and gathering recognition in dormitories, and climbing and area intrusion recognition at perimeter walls and hazardous areas. Keep the rest for now and enable them at any time if needed.

Q2: Do the algorithms run locally or in the cloud? Will videos leave the school?

A: By default, they run locally. SEEMSE (Jinzhang Technology) edge servers are placed in the school computer room. Camera images come in and are analyzed on site. Only recognition results and alarms go out; raw videos do not need to be uploaded to the cloud, greatly reducing bandwidth and privacy risks.

Q3: Once an alarm is reported, how is it delivered to the homeroom teacher as soon as possible?

A: Three channels work together: mobile app message push, desktop reminder pop-up, and SMS as a fallback. Recipients can also be designated by point: dormitory alarms are sent to dormitory managers, school gate alarms to the security office, and homeroom teachers receive only those related to their own classes.

Q4: Can the system detect things like campus bullying in advance?

A: It can detect signs, but it cannot predict. Once behaviors such as fighting, gathering and crowding around, or smoking occur, the algorithm alarms on the spot, much faster than reviewing surveillance afterward. Truly suppressing them still requires cooperation with the school's management measures; the system is responsible for letting teachers know earlier.

Q5: Can one system manage multiple campuses together?

A: Yes. Each campus has one streaming media server and one edge server, with images processed nearby and then uniformly aggregated to the Web-based management platform. On-campus traffic uses the intranet, and cross-campus traffic is forwarded over the public network, so devices and alarms from all campuses can be seen in one backend.