Skip to main content
Version: Next

Chat

EVA provides an innovative feature that allows users to directly control the system by entering natural language commands through the chat interface provided for each camera.

When users input their requests in natural language, the LLM (Large Language Model) in the EVA Agent understands them and controls various EVA functions in real-time.

Through interaction with EVA, you can apply AI analysis settings for cameras more effectively and intuitively.



Object Detection Settings Control​

  • Set/Modify Target Objects:

    "Set the detection targets to 'person' and 'dog'."

  • Adjust Threshold:

    "Set the threshold for car detection to 0.85." Increase detection confidence to detect only certain objects or lower it to adjust sensitivity.

  • Initiate Few-Shot Learning:

    "Let's start few-shot learning for 'my cat'." Start learning for new objects to enable image-guided detection.



Alarm and Scenario Control​

  • Modify Alarm Scenario:

    "Notify me if a fire breaks out in the mountains."

  • Adjust Alarm Interval:

    "Change the alert interval to 10." Adjust the interval for checking alarms if they are too frequent or too delayed.

  • Adjust Feedback Similarity Threshold:

    "Change the feedback similarity threshold to 0.95." EVA sends alarms based on previously received feedback. Raising the threshold helps prevent unnecessary alarms, while lowering it helps avoid missing alarms.