Use of object store memory

How severe does this issue affect your experience of using Ray?

  • Medium: It contributes to significant difficulty to complete my task, but I can work around it.

I have three modules in my ray pipeline: camera, detection and tracker. There is frame_queue (ray queue) to share frames from camera module to detection module and det_queue (ray queue) to share detections from detection module to tracker module. Now, in camera module, I am reading rtsp streaming using cv2, creating a instance of Frame class ( mentioned below) with only frame, and pushing it to frame_queue. In detection module, I am poping instance, getting detection for frame, updating the same instances with dets and pushing same instance in det_queue.

In this flow, in camera module when we are passing a instance in frame_queue, it will upload it to object store memory and share reference of it into queue and in detection module it will read instance from object store memory again. But now, when we update the instance, will it upload the complete instance into object store memory or update the already existed instance?

Please let me know, if I am missing anything.

class Frame:
    def __init__(self, frame, dets=None):
        self.frame = frame
        self.dets = dets

hi @shyampatel
can you show a bit more code on how you use ray queue to update the frames?