583
edits
DeepStream Reference Designs/Project Architecture/High Level Design: Difference between revisions
DeepStream Reference Designs/Project Architecture/High Level Design (view source)
Revision as of 23:42, 9 May 2022
, 9 May 2022→Custom Application Specific
| Line 26: | Line 26: | ||
For instance, take a parking lot application. In that scenario, you could use RTSP cameras and perform inference over the captured images and receive the predictions through the MQTT message broker. These notifications trigger the Parking Lot Business Rules which evaluate the current system state and decide if the different actions should be triggered. | For instance, take a parking lot application. In that scenario, you could use RTSP cameras and perform inference over the captured images and receive the predictions through the MQTT message broker. These notifications trigger the Parking Lot Business Rules which evaluate the current system state and decide if the different actions should be triggered. | ||
Now imagine you wanted to implement a shoplifting detection system. This time a GigE camera is used, however, no modification is needed in the camera capture module since it is protected by the interface. Predictions are made using a different model, which is then sent to the Shoplifting Business Rules. | Now imagine you wanted to implement a shoplifting detection system. This time a GigE camera is used, however, no modification is needed in the camera capture module since it is protected by the interface. Predictions are made using a different model, which is then sent to the Shoplifting Business Rules. As you can see, there are a lot of possibilities to leverage this reference design as a starting point for your specific project. | ||