Detection Scenario
The first step in starting AI analysis with the EVA system is to create a Detection Scenario.
This scenario goes beyond merely detecting the presence of objects, defining natural language-based rules that guide the AI to evaluate situations and provide alarms only under desired conditions.
Refining Scenarios
The more detailed a Detection Scenario is, the more accurately the EVA Agent understands the situation.
To reduce false alarms, it is crucial to clearly specify situations to exclude or specific exception conditions.
Scenario Creation Support with the Generate Function
To assist users in creating scenarios, EVA provides the Generate function.
- Enter Draft: Input a brief draft of the desired situation.
- Click Generate: Clicking the button prompts the EVA Agent’s LLM to expand and provide a refined template based on the provided information.
- Refinement: Users can further polish the scenario using the generated template for greater precision.
💡 When using the Generate function, EVA utilizes the camera’s images and meta information (e.g., installation location, operational purpose). Providing detailed information about the purpose and location of the camera enhances the Agent’s understanding of the situation, improving the accuracy of scenario expansion.
Situation Analysis
The finalized detection scenario is sent to the EVA Agent along with an image snapshot when specific conditions are met for detected objects. The EVA Agent uses VLM to analyze the appropriateness of the situation based on the scenario, makes a final decision on whether to trigger an alarm, and sends a detailed message to the user.