This article explores the benefits offered by natural language rules engines in video surveillance.
Setting up alarms using natural language is a fascinating topic that is gaining momentum in video surveillance. This can be achieved with natural language-based rules engines, which simplify the alarm creation process so that practically anyone, not just engineers or professionals, can create algorithms and alarms. This article explores the benefits offered by natural language rules engines in video surveillance.
Imagine you want to make your video surveillance system so that it issues an alarm when a car stops in the loading zone for more than 5 minutes. In conventional rules engines, this can be a quite complex process as it entails the following: material collection, data annotation, algorithm training and deployment on device, done by people who are professionally trained.
Now imagine if you can just input the text – or even say – “Alert me if a car stops in the loading zone for more than 5 minutes” into the system, and it creates the algorithm for you. This is where natural language-based rules engines come in. A natural language-based rules engine is a rules engine that allows users to define surveillance rules in plain language – for example “Alert me if a person enters the restricted zone after 10 PM.” The technology has generated much interest in video surveillance, amid the emergence of natural language models, large-scale AI models and ChatGPT.
Benefits of natural language-based rules engines include the following:
Ease of use: While conventional rules engines need complex configuration or scripting, natural language makes it accessible to non-technical users; there’s no need for programming or low-level configuration;
Flexibility: Natural language-based rules engines allow users to easily update or add new rules without retraining the system;
Scalability: Algorithms developed as such are deployable across multiple sites and cameras without heavy manual setup;
Dynamic control: Security teams can change rules in real-time via natural language commands.
“Natural language-based engines make video surveillance more accessible by allowing users to set rules in plain English, reducing technical barriers. This leads to faster, more user-friendly searches and forensic reviews – especially when combined with strong computer vision in hybrid edge-cloud setups,” said Albert Stepanyan, President and CEO of Scylla AI.
“Conventional systems often require manual zone drawing and technical parameter settings through complex interfaces. Natural language-based engines simplify this by enabling intuitive setup and more accurate, user-friendly rule creation, without requiring deep technical expertise,” he adds.
Stepanyan continues: “Verticals like healthcare, education, retail, and critical infrastructure stand to benefit the most, since non-technical staff can efficiently configure and manage surveillance without specialized training.”
Core components needed for natural language rules engines
Natural language-based rules engines work by way of natural language understanding, which entails intent detection, such as “detect entry,” “track object,” and “send alert”; and entity extraction, which identifies objects (person, car), actions (entering, loitering) and conditions (time, location). Rule parsing and mapping then takes place, where natural language command is converted into a structured rule such as “Alert if a person enters Zone A after 8 PM.” The event trigger and response mechanism will then be set, where what happens when rule conditions are met will be defined.
To ensure the above can be executed properly, several core components are needed. These include NLP (natural language processing) models, computer vision models, rules engine and an integration layer connecting NLU outputs with video analytics and alert systems.
“Key components include computer vision models for object detection, hybrid edge-cloud processing for real-time performance, and lightweight NLP for interpreting queries. The emphasis is on efficient vision-based analytics, with NLP serving as a bridge to make interaction natural and reliable,” Stepanyan said.
Product Adopted:Others