
School Security Management Challenges
Incivility a Challenge: Difficult to PreventSmoking and other uncivil behaviors in certain areas on campus pose significant challenges in locating individuals, making effective prevention difficult.
Unsafe behavior, difficult to control:There is a lack of effective detection methods for unsafe behaviors on campus, such as carrying passengers on bicycles or using mobile phones while riding, which hinders the formation of a linkage and control mechanism.
Lost and hard to find, service is difficult:Items and children going missing occur frequently, making it difficult to locate and trace, and hard to quickly pinpoint targets for timely service.
When school security management challenges meet Hikvision Search:
Hikvision's Wen Sou product, integrated with large model technology, further empowers campus safety management and enhances video value.By leveraging the Wensou mainframe, we achieve comprehensive search and text alert capabilities across the campus, while fine-tuning the model to enhance its accuracy over time. Through the application and exploration of large-scale models, we better safeguard campus safety and order, providing more precise and warm services to both students and faculty.

Wen Sou Application Architecture

Core Features and Typical Scenarios
Riding bicycles with passengers: Immediate warnings for unsafe practices
Guarding the safety of students and teachers
There's a high number of people riding electric scooters on university campuses, often accompanied by unsafe behaviors such as carrying passengers and not wearing helmets. In response, Wen Sou Host supports text-based warnings, utilizing the Guanlan large model to add positive and negative prompts in text descriptions. This integration with the campus platform facilitates SMS reminders for such behaviors.

Text Alert:By describing the early warning rules through text, it offers simplicity, ease of use, and efficiency, compatible with various complex and long-tail scenarios for early warning. It achieves multi-modal early warning and allows for continuous optimization.
Multi-scenario Universal Search
Quickly locate lost items and missing individuals
The university campus boasts a rich academic atmosphere and cultural heritage, often attracting visitors and students from other schools. In cases of missing individuals, the Wensou mainframe can integrate large model technology to quickly locate the missing person based on their physical features. Similarly, by identifying item characteristics, lost items can also be swiftly located, assisting owners in retrieving their belongings.

Image Search by TextUtilizing large model technology, we have achieved high-dimensional vector representation of candid photos, enabling open-ended text-based image search for people, vehicles, objects, events, and locations.
Utilizing legacy non-intelligent devices to achieve multi-dimensional perception
Comprehensive Protection of Campus Safety
Utilizing the existing non-intelligent IPCs and other sensory devices within the old campus, we are exploring more intelligent applications, significantly enhancing recognition accuracy, and comprehensively safeguarding campus security.

Campus Security AI Model Becomes More Accurate with Use
Crafting Unique Intelligent Security Construction Resources
Schools can leverage the universal search capability of WenSousu products to explore more intelligent applications. Based on the "search by text to find images" results from the WenSousu host, the search results are sorted by accuracy and continuously trained and optimized for better performance over time.

✔Model Fine-tuning:After the initial search, the model is fine-tuned, and data selections are made correctly or incorrectly to optimize the effects.
Featured Case: Wuhan University
Wuhan University, spanning...Luo Jia MountainThe century-old university by the East Lake, with its expansive space, high population density, and complex scenarios, presents a "multidimensional equation" challenge to traditional security systems.
Hikvision offers a Search Host Application Solution, enhancing the security view value of Wuhan University's Luojia Mountain Campus, creating an integrated, intelligent, and visual video intelligence campus platform.
This project implements intelligent textual analysis on the school's regular surveillance equipment, connects with the campus platform and operator systems, to promptly alert against unsafe or uncivil behaviors such as cherry blossom picking and carrying passengers; swiftly locate lost children and missing items; and actively prevent risky activities like distributing flyers and smoking on construction sites, effectively enhancing the school's safety and management防范.





