How to include non-continuous stream / camera in trigger mode source in Deepstream pipeline?

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU): Jetson AGX Xavier
• DeepStream Version: 5.0
• JetPack Version (valid for Jetson only): 4.4
• TensorRT Version: 7.1.3
• Issue Type( questions, new requirements, bugs): question

Hi DS team,

I have an application where the live capture source (a Basler camera) doesn’t stream continuously, it waits for a trigger -> starts capture images -> goes back to idle -> waits for a trigger -> [repeat cycle]

The idle duration is not fixed since it depends on the time taken for the item to reaches the trigger point. If I understand correctly, the deepstream-app and test-apps don’t currently support this use case? If so would you please share any example or advice on how to use camera in trigger mode as source?

What type of software has the camera manufacture provided you? Is the camera driver based on v4l2 interface?
The DeepStream is a framework/middleware level application. It only interacts with devices through some standard interfaces such as v4l2. The specific camera operation should be implemented in library or driver level, it has nothing to do with middleware.

Hi @Fiona.Chen,

My camera is a Basler acA2040-55uc. It doesn’t support the v4l2 interface. From its specifications, I think uses USB3 Vision interface.

I already found a plugin that works with the camera, however, the camera doesn’t capture images continuously, it operates in trigger mode, e.g., the camera stays idle and wakes up whenever an item moves cross the sensor. I want to ask if I need to take care of source that non-continuous stream in some special way.

USB3 interface is hardware interface, we are talking about the driver and software interface.https://www.kernel.org/doc/html/v4.9/media/uapi/v4l/v4l2.html

You don’t need to so anything with DeepStream. All control related things should be done within the driver and middleware.