Add Frame to MSGBroker payload using EventMetaData

#!/usr/bin/env python3

# SPDX-FileCopyrightText: Copyright (c) 2019-2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# See the License for the specific language governing permissions and
# limitations under the License.

import sys

import gi

gi.require_version('Gst', '1.0')
from gi.repository import GLib, Gst
import sys
from optparse import OptionParser
from common.is_aarch_64 import is_aarch64
from common.bus_call import bus_call
from common.utils import long_to_uint64
import pyds
import argparse
from common.FPS import PERF_DATA

input_file = None
cfg_file = None
topic = None
no_display = False
perf_data = None

GPU_ID = 0

PGIE_CONFIG_FILE = "dstest4_pgie_config.txt"
MSCONV_CONFIG_FILE = "dstest4_msgconv_config.txt"

pgie_classes_str = ["Vehicle", "TwoWheeler", "Person", "Roadsign"]

def set_pgie_properties(pgie, pgie_config_path, gpu_id, number_sources):
    pgie.set_property('config-file-path', PGIE_CONFIG_FILE)
    pgie.set_property('gpu_id', gpu_id)
    pgie_batch_size = pgie.get_property("batch-size")
    if pgie_batch_size != number_sources:
            "WARNING: Overriding infer-config batch-size",
            " with number of sources ",
            " \n",
        pgie.set_property("batch-size", number_sources)

def cb_newpad(decodebin, decoder_src_pad, data):
    The function is called when a new pad is created by the decodebin.
    The function checks if the new pad is for video and not audio.
    If the new pad is for video, the function checks if the pad caps contain NVMM memory features.
    If the pad caps contain NVMM memory features, the function links the decodebin pad to the source bin
    ghost pad.
    If the pad caps do not contain NVMM memory features, the function prints an error message.
    :param decodebin: The decodebin element that is creating the new pad
    :param decoder_src_pad: The source pad created by the decodebin element
    :param data: This is the data that was passed to the callback function. In this case, it is the
    print("In cb_newpad\n")
    caps = decoder_src_pad.get_current_caps()
    gststruct = caps.get_structure(0)
    gstname = gststruct.get_name()
    source_bin = data
    features = caps.get_features(0)

    # Need to check if the pad created by the decodebin is for video and not
    # audio.
    print("gstname=", gstname)
    if gstname.find("video") != -1:
        # Link the decodebin pad only if decodebin has picked nvidia
        # decoder plugin nvdec_*. We do this by checking if the pad caps contain
        # NVMM memory features.
        print("features=", features)
        if features.contains("memory:NVMM"):
            # Get the source bin ghost pad
            bin_ghost_pad = source_bin.get_static_pad("src")
            if not bin_ghost_pad.set_target(decoder_src_pad):
                    "Failed to link decoder src pad to source bin ghost pad\n"
            sys.stderr.write(" Error: Decodebin did not pick nvidia decoder plugin.\n")

def decodebin_child_added(child_proxy, Object, name, user_data):
    If the child added to the decodebin is another decodebin, connect to its child-added signal. If the
    child added is a source, set its drop-on-latency property to True.

    :param child_proxy: The child element that was added to the decodebin
    :param Object: The object that emitted the signal
    :param name: The name of the element that was added
    :param user_data: This is a pointer to the data that you want to pass to the callback function
    print("Decodebin child added:", name, "\n")
    if name.find("decodebin") != -1:
        Object.connect("child-added", decodebin_child_added, user_data)

    if "source" in name:
        source_element = child_proxy.get_by_name("source")
        if source_element.find_property("drop-on-latency") != None:
            Object.set_property("drop-on-latency", True)

def create_source_bin(index, uri):
    It creates a GstBin, adds a uridecodebin to it, and connects the uridecodebin's pad-added signal to
    a callback function

    :param index: The index of the source bin
    :param uri: The URI of the video file to be played
    :return: A bin with a uri decode bin and a ghost pad.
    print("Creating source bin")

    # Create a source GstBin to abstract this bin's content from the rest of the
    # pipeline
    bin_name = "source-bin-%02d" % index
    nbin =
    if not nbin:
        sys.stderr.write(" Unable to create source bin \n")

    # Source element for reading from the uri.
    # We will use decodebin and let it figure out the container format of the
    # stream and the codec and plug the appropriate demux and decode plugins.
    uri_decode_bin = Gst.ElementFactory.make("uridecodebin", "uri-decode-bin")
    if not uri_decode_bin:
        sys.stderr.write(" Unable to create uri decode bin \n")
    # We set the input uri to the source element
    uri_decode_bin.set_property("uri", uri)
    # Connect to the "pad-added" signal of the decodebin which generates a
    # callback once a new pad for raw data has beed created by the decodebin
    uri_decode_bin.connect("pad-added", cb_newpad, nbin)
    uri_decode_bin.connect("child-added", decodebin_child_added, nbin)

    # We need to create a ghost pad for the source bin which will act as a proxy
    # for the video decoder src pad. The ghost pad will not have a target right
    # now. Once the decode bin creates the video decoder and generates the
    # cb_newpad callback, we will set the ghost pad target to the video decoder
    # src pad.
    Gst.Bin.add(nbin, uri_decode_bin)
    bin_pad = nbin.add_pad(Gst.GhostPad.new_no_target("src", Gst.PadDirection.SRC))
    if not bin_pad:
        sys.stderr.write(" Failed to add ghost pad in source bin \n")
        return None
    return nbin

def make_element(element_name, i):
    Creates a Gstreamer element with unique name
    Unique name is created by adding element type and index e.g. `element_name-i`
    Unique name is essential for all the element in pipeline otherwise gstreamer will throw exception.
    :param element_name: The name of the element to create
    :param i: the index of the element in the pipeline
    :return: A Gst.Element object
    element = Gst.ElementFactory.make(element_name, element_name)
    if not element:
        sys.stderr.write(" Unable to create {0}".format(element_name))
    element.set_property("name", "{0}-{1}".format(element_name, str(i)))
    return element

# Callback function for deep-copying an NvDsEventMsgMeta struct
def meta_copy_func(data, user_data):
    # Cast data to pyds.NvDsUserMeta
    user_meta = pyds.NvDsUserMeta.cast(data)
    src_meta_data = user_meta.user_meta_data
    # Cast src_meta_data to pyds.NvDsEventMsgMeta
    srcmeta = pyds.NvDsEventMsgMeta.cast(src_meta_data)
    # Duplicate the memory contents of srcmeta to dstmeta
    # First use pyds.get_ptr() to get the C address of srcmeta, then
    # use pyds.memdup() to allocate dstmeta and copy srcmeta into it.
    # pyds.memdup returns C address of the allocated duplicate.
    dstmeta_ptr = pyds.memdup(pyds.get_ptr(srcmeta),
    # Cast the duplicated memory to pyds.NvDsEventMsgMeta
    dstmeta = pyds.NvDsEventMsgMeta.cast(dstmeta_ptr)

    # Duplicate contents of ts field. Note that reading srcmeat.ts
    # returns its C address. This allows to memory operations to be
    # performed on it.
    dstmeta.ts = pyds.memdup(srcmeta.ts, MAX_TIME_STAMP_LEN + 1)

    # Copy the sensorStr. This field is a string property. The getter (read)
    # returns its C address. The setter (write) takes string as input,
    # allocates a string buffer and copies the input string into it.
    # pyds.get_string() takes C address of a string and returns the reference
    # to a string object and the assignment inside the binder copies content.
    dstmeta.sensorStr = pyds.get_string(srcmeta.sensorStr)

    if srcmeta.objSignature.size > 0:
        dstmeta.objSignature.signature = pyds.memdup(
            srcmeta.objSignature.signature, srcmeta.objSignature.size)
        dstmeta.objSignature.size = srcmeta.objSignature.size

    if srcmeta.extMsgSize > 0:
        if srcmeta.objType == pyds.NvDsObjectType.NVDS_OBJECT_TYPE_VEHICLE:
            srcobj = pyds.NvDsVehicleObject.cast(srcmeta.extMsg)
            obj = pyds.alloc_nvds_vehicle_object()
            obj.type = pyds.get_string(srcobj.type)
            obj.make = pyds.get_string(srcobj.make)
            obj.model = pyds.get_string(srcobj.model)
            obj.color = pyds.get_string(srcobj.color)
            obj.license = pyds.get_string(srcobj.license)
            obj.region = pyds.get_string(srcobj.region)
            dstmeta.extMsg = obj
            dstmeta.extMsgSize = sys.getsizeof(pyds.NvDsVehicleObject)
        if srcmeta.objType == pyds.NvDsObjectType.NVDS_OBJECT_TYPE_PERSON:
            srcobj = pyds.NvDsPersonObject.cast(srcmeta.extMsg)
            obj = pyds.alloc_nvds_person_object()
            obj.age = srcobj.age
            obj.gender = pyds.get_string(srcobj.gender)
            obj.cap = pyds.get_string(srcobj.cap)
   = pyds.get_string(
            obj.apparel = pyds.get_string(srcobj.apparel)
            dstmeta.extMsg = obj
            dstmeta.extMsgSize = sys.getsizeof(pyds.NvDsVehicleObject)

    return dstmeta

# Callback function for freeing an NvDsEventMsgMeta instance
def meta_free_func(data, user_data):
    user_meta = pyds.NvDsUserMeta.cast(data)
    srcmeta = pyds.NvDsEventMsgMeta.cast(user_meta.user_meta_data)

    # pyds.free_buffer takes C address of a buffer and frees the memory
    # It's a NOP if the address is NULL

    if srcmeta.objSignature.size > 0:
        srcmeta.objSignature.size = 0

    if srcmeta.extMsgSize > 0:
        if srcmeta.objType == pyds.NvDsObjectType.NVDS_OBJECT_TYPE_VEHICLE:
            obj = pyds.NvDsVehicleObject.cast(srcmeta.extMsg)
        if srcmeta.objType == pyds.NvDsObjectType.NVDS_OBJECT_TYPE_PERSON:
            obj = pyds.NvDsPersonObject.cast(srcmeta.extMsg)
        srcmeta.extMsgSize = 0

def generate_vehicle_meta(data):
    obj = pyds.NvDsVehicleObject.cast(data)
    obj.type = "sedan"
    obj.color = "blue"
    obj.make = "Bugatti"
    obj.model = "M"
    obj.license = "XX1234"
    obj.region = "CA"
    return obj

def generate_person_meta(data):
    obj = pyds.NvDsPersonObject.cast(data)
    obj.age = 45
    obj.cap = "none" = "black"
    obj.gender = "male"
    obj.apparel = "formal"
    return obj

def generate_event_msg_meta(data, class_id):
    meta = pyds.NvDsEventMsgMeta.cast(data)
    meta.sensorId = 0
    meta.placeId = 0
    meta.moduleId = 0
    meta.sensorStr = "sensor-0"
    meta.ts = pyds.alloc_buffer(MAX_TIME_STAMP_LEN + 1)
    pyds.generate_ts_rfc3339(meta.ts, MAX_TIME_STAMP_LEN)

    # This demonstrates how to attach custom objects.
    # Any custom object as per requirement can be generated and attached
    # like NvDsVehicleObject / NvDsPersonObject. Then that object should
    # be handled in payload generator library (nvmsgconv.cpp) accordingly.
    if class_id == PGIE_CLASS_ID_VEHICLE:
        meta.type = pyds.NvDsEventType.NVDS_EVENT_MOVING
        meta.objType = pyds.NvDsObjectType.NVDS_OBJECT_TYPE_VEHICLE
        meta.objClassId = PGIE_CLASS_ID_VEHICLE
        obj = pyds.alloc_nvds_vehicle_object()
        obj = generate_vehicle_meta(obj)
        meta.extMsg = obj
        meta.extMsgSize = sys.getsizeof(pyds.NvDsVehicleObject)
    if class_id == PGIE_CLASS_ID_PERSON:
        meta.type = pyds.NvDsEventType.NVDS_EVENT_ENTRY
        meta.objType = pyds.NvDsObjectType.NVDS_OBJECT_TYPE_PERSON
        meta.objClassId = PGIE_CLASS_ID_PERSON
        obj = pyds.alloc_nvds_person_object()
        obj = generate_person_meta(obj)
        meta.extMsg = obj
        meta.extMsgSize = sys.getsizeof(pyds.NvDsPersonObject)
    return meta

# osd_sink_pad_buffer_probe  will extract metadata received on OSD sink pad
# and update params for drawing rectangle, object information etc.
# a) probe() callbacks are synchronous and thus holds the buffer
#    (info.get_buffer()) from traversing the pipeline until user return.
# b) loops inside probe() callback could be costly in python.
#    So users shall optimize according to their use-case.
def osd_sink_pad_buffer_probe(pad, info, u_data):
    frame_number = 0
    # Intiallizing object counter with 0.
    obj_counter = {
    gst_buffer = info.get_buffer()
    if not gst_buffer:
        print("Unable to get GstBuffer ")

    # Retrieve batch metadata from the gst_buffer
    # Note that pyds.gst_buffer_get_nvds_batch_meta() expects the
    # C address of gst_buffer as input, which is obtained with hash(gst_buffer)
    batch_meta = pyds.gst_buffer_get_nvds_batch_meta(hash(gst_buffer))
    if not batch_meta:
        return Gst.PadProbeReturn.OK
    l_frame = batch_meta.frame_meta_list
    while l_frame is not None:
            # Note that needs a cast to pyds.NvDsFrameMeta
            # The casting is done by pyds.NvDsFrameMeta.cast()
            # The casting also keeps ownership of the underlying memory
            # in the C code, so the Python garbage collector will leave
            # it alone.
            frame_meta = pyds.NvDsFrameMeta.cast(
        except StopIteration:
        is_first_object = True

        # Short example of attribute access for frame_meta:
        # print("Frame Number is ", frame_meta.frame_num)
        # print("Source id is ", frame_meta.source_id)
        # print("Batch id is ", frame_meta.batch_id)
        # print("Source Frame Width ", frame_meta.source_frame_width)
        # print("Source Frame Height ", frame_meta.source_frame_height)
        # print("Num object meta ", frame_meta.num_obj_meta)

        frame_number = frame_meta.frame_num
        l_obj = frame_meta.obj_meta_list
        while l_obj is not None:
                obj_meta = pyds.NvDsObjectMeta.cast(
            except StopIteration:

            # Update the object text display
            txt_params = obj_meta.text_params

            # Set display_text. Any existing display_text string will be
            # freed by the bindings module.
            txt_params.display_text = pgie_classes_str[obj_meta.class_id]

            obj_counter[obj_meta.class_id] += 1

            # Font , font-color and font-size
            txt_params.font_params.font_name = "Serif"
            txt_params.font_params.font_size = 10
            # set(red, green, blue, alpha); set to White
            txt_params.font_params.font_color.set(1.0, 1.0, 1.0, 1.0)

            # Text background color
            txt_params.set_bg_clr = 1
            # set(red, green, blue, alpha); set to Black
            txt_params.text_bg_clr.set(0.0, 0.0, 0.0, 1.0)

            # Ideally NVDS_EVENT_MSG_META should be attached to buffer by the
            # component implementing detection / recognition logic.
            # Here it demonstrates how to use / attach that meta data.
            if (frame_number % 2) == 0:
                # Frequency of messages to be send will be based on use case.
                # Here message is being sent for first object every 30 frames.

                # Allocating an NvDsEventMsgMeta instance and getting
                # reference to it. The underlying memory is not manged by
                # Python so that downstream plugins can access it. Otherwise
                # the garbage collector will free it when this probe exits.
                msg_meta = pyds.alloc_nvds_event_msg_meta()
                msg_meta.bbox.left = obj_meta.rect_params.left
                msg_meta.bbox.width = obj_meta.rect_params.width
                msg_meta.bbox.height = obj_meta.rect_params.height
                msg_meta.frameId = frame_number
                msg_meta.trackingId = long_to_uint64(obj_meta.object_id)
                msg_meta.confidence = obj_meta.confidence
                msg_meta = generate_event_msg_meta(msg_meta, obj_meta.class_id)
                user_event_meta = pyds.nvds_acquire_user_meta_from_pool(
                if user_event_meta:
                    user_event_meta.user_meta_data = msg_meta
                    user_event_meta.base_meta.meta_type = pyds.NvDsMetaType.NVDS_EVENT_MSG_META
                    # Setting callbacks in the event msg meta. The bindings
                    # layer will wrap these callables in C functions.
                    # Currently only one set of callbacks is supported.
                    pyds.user_copyfunc(user_event_meta, meta_copy_func)
                    pyds.user_releasefunc(user_event_meta, meta_free_func)
                    print("Error in attaching event meta to buffer\n")

                is_first_object = False
                l_obj =
            except StopIteration:
            l_frame =
        except StopIteration:

    print("Frame Number =", frame_number, "Vehicle Count =",
          obj_counter[PGIE_CLASS_ID_VEHICLE], "Person Count =",
    return Gst.PadProbeReturn.OK

def main(args):
    # registering callbacks

    input_sources = args.input_videos
    number_sources = len(input_sources)
    global perf_data
    perf_data = PERF_DATA(number_sources)

    pgie_config_path = args.pgie_config_path

    # Standard GStreamer initialization

    # Create gstreamer elements */
    # Create Pipeline element that will form a connection of other elements
    print("Creating Pipeline \n ")
    pipeline = Gst.Pipeline()
    is_live = False

    if not pipeline:
        sys.stderr.write("Unable to create Pipeline \n")

    print("Creating streamux \n ")

    # Create nvstreammux instance to form batches from one or more sources.
    streammux = Gst.ElementFactory.make("nvstreammux", "Stream-muxer")
    if not streammux:
        sys.stderr.write(" Unable to create NvStreamMux \n")

    for i in range(number_sources):
        print("Creating source_bin ", i, " \n ")
        uri_name = input_sources[i]
        if uri_name.find("rtsp://") == 0:
            is_live = True
        source_bin = create_source_bin(i, uri_name)
        if not source_bin:
            sys.stderr.write("Unable to create source bin \n")
        padname = "sink_%u" % i
        sinkpad = streammux.get_request_pad(padname)
        if not sinkpad:
            sys.stderr.write("Unable to create sink pad bin \n")
        srcpad = source_bin.get_static_pad("src")
        if not srcpad:
            sys.stderr.write("Unable to create src pad bin \n")

    queue1 = Gst.ElementFactory.make("queue", "queue1")
    print("Creating Pgie \n ")
    pgie = Gst.ElementFactory.make("nvinfer", "primary-inference")
    if not pgie:
        sys.stderr.write(" Unable to create pgie \n")

    print("Creating nvstreamdemux \n ")
    nvstreamdemux = Gst.ElementFactory.make("nvstreamdemux", "nvstreamdemux")
    if not nvstreamdemux:
        sys.stderr.write(" Unable to create nvstreamdemux \n")

    if is_live:
        print("Atleast one of the sources is live")
        streammux.set_property("live-source", 1)

    streammux.set_property('width', 1920)
    streammux.set_property('height', 1080)
    streammux.set_property("batch-size", number_sources)
    streammux.set_property("batched-push-timeout", 4000000)
    set_pgie_properties(pgie, pgie_config_path, GPU_ID, number_sources)
    pgie_batch_size = pgie.get_property("batch-size")
    if pgie_batch_size != number_sources:
            "WARNING: Overriding infer-config batch-size",
            " with number of sources ",
            " \n",
        pgie.set_property("batch-size", number_sources)

    print("Adding elements to Pipeline \n")

    # linking
    ##creating demux src

    for i in range(number_sources):
        print("Creating FakeSink \n")
        sink = make_element("fakesink", i)
        if not sink:
            sys.stderr.write(" Unable to create fakesink \n")

        # creating queue
        queue = make_element("queue", i)

        # creating nvvidconv
        nvvideoconvert = make_element("nvvideoconvert", i)

        # creating nvosd
        nvdsosd = make_element("nvdsosd", i)
        nvdsosd.set_property("process-mode", OSD_PROCESS_MODE)
        nvdsosd.set_property("display-text", OSD_DISPLAY_TEXT)

        # Creating tee to split msgbroker + sink
        tee = make_element("tee", i)
        sink_queue = make_element("queue", f"sink-{i}")
        broker_queue = make_element("queue", f"broker-{i}")

        msgbroker = make_element("nvmsgbroker", i)
        msgconv = make_element("nvmsgconv", i)

        msgconv.set_property('config', args.msgconv_config_path)
        msgconv.set_property('payload-type', 1)
        msgconv.set_property('comp-id', i)
        #msgconv.set_property('config', MSCONV_CONFIG_FILE)

        msgbroker.set_property('proto-lib', "/opt/nvidia/deepstream/deepstream/lib/")
        msgbroker.set_property('config', args.msgbroker_config_path)
        msgbroker.set_property('sync', False)

        topic = f"event.{i}"
        msgbroker.set_property('topic', topic)
        msgbroker.set_property('comp-id', i)


        # connect nvstreamdemux -> queue
        padname = "src_%u" % i
        demuxsrcpad = nvstreamdemux.get_request_pad(padname)
        if not demuxsrcpad:
            sys.stderr.write("Unable to create demux src pad \n")

        queuesinkpad = queue.get_static_pad("sink")
        if not queuesinkpad:
            sys.stderr.write("Unable to create queue sink pad \n")

        # connect tee -> broker_queue/sink_queue
        sink_pad = broker_queue.get_static_pad("sink")
        tee_msg_pad = tee.get_request_pad('src_%u')
        tee_render_pad = tee.get_request_pad("src_%u")
        if not tee_msg_pad or not tee_render_pad:
            sys.stderr.write("Unable to get request pads\n")
        sink_pad = sink_queue.get_static_pad("sink")

        # connect  queue -> nvvidconv -> nvosd -> tee -> broker_queue/sink_queue

        sink.set_property("qos", 0)

        # Lets add probe to get informed of the meta data generated, we add probe to
        # the sink pad of the osd element, since by that time, the buffer would have
        # had got all the metadata.
        osdsinkpad = nvdsosd.get_static_pad("sink")
        if not osdsinkpad:
            sys.stderr.write(" Unable to get sink pad of nvosd \n")

        osdsinkpad.add_probe(Gst.PadProbeType.BUFFER, osd_sink_pad_buffer_probe, 0)

    print("Linking elements in the Pipeline \n")
    # create an event loop and feed gstreamer bus mesages to it
    loop = GLib.MainLoop()
    bus = pipeline.get_bus()
    bus.connect("message", bus_call, loop)

    # List the sources
    print("Now playing...")
    for i, source in enumerate(input_sources):
        print(i, ": ", source)

    print("Starting pipeline \n")
    # start play back and listed to events

    # cleanup
    print("Exiting app\n")

# Parse and validate input arguments
def parse_args():
    parser = argparse.ArgumentParser(description='')
                        nargs = '+',
                        help='Base topic to be used to send payloads')

    args = parser.parse_args()
    return args

if __name__ == '__main__':
    ret = parse_args()

I ran it by putting it in the deepstream_test_4 directory and running the command below:

python3 --topic abc --input-videos file:///opt/nvidia/deepstream/deepstream-6.3/samples/streams/sample_1080p_h264.mp4 file:///opt/nvidia/deepstream/deepstream-6.3/samples/streams/sample_1080p_h265.mp4 --pgie-config-path dstest4_pgie_config.txt --msgconv-config-path dstest4_msgconv_config.txt --msgbroker-config-path cfg_amqp.txt

This is essentially a combination of deepstream_test_4 and multi_in_multi_out.

Is this still an DeepStream issue to support? Thanks!

I posted a test script and steps to reproduce the issue above.

I modified the code to send kafka. I can 't reproduce the crash problem, and the kafka sever can receive the messages.
test-log.txt (163.7 KB)
kafka_server.txt (1.5 MB) (28.1 KB)

Were you able to reproduce it with AMQP?

  1. after testing kafka or AMQP sending again, the app crashed sometimes. testing without any modification, the app sent AMQP fine.
  2. notice you are using two sperate nvmsgconv+nvmsgborker, which has own componentId, but you did not set componentId of NvDsEventMsgMeta correctly. please refer to part 9 in opt\nvidia\deepstream\deepstream\sources\apps\sample_apps\deepstream-test5\README.
  3. why do you need nvstreamdemux to use two sperate nvmsgconv+nvmsgborker? if using one nvmsgconv+nvmsgborker, it will be similar as

For number 2, that’s very interesting. Did the crash go away when you corrected that?

Looking at our app(not the one I sent you), it looks like we do set that value in generate_event_msg_meta. It’s missing from our meta_copy_func though. Does it need to be? I see that things like meta.moduleId and meta.sensorId aren’t copied or freed in those two functions.

meta.componentId = index

For number 3, the demuxer is used because we have multiple streams that run through the same model, but need slightly different behavior based on which stream it is. We want different streams to be sent to different topics, and slightly different processing to be done in the probe. We also need to save the frames to a video, as this will eventually be coming from an RTSP stream, so I do think we need to demux.

Can we change the topic based on stream without having multiple msg brokers? Or should we have only one msgconv with two msg brokers? With the needs mentioned above, what structure would you recommend?

this value is needed. nvmsgcov and nvmsgbroker will check this value.

if using multiple topics, multiple nvmsgconv+nvmsgbroker are needed. please refer to this topic. to narrow down this issue, we will use C code to reproduce this issue.

  1. Based on your code, I modified “nvsmsgconv + nvmsgbroker” to fakesink, but the app will hang. did you get the same result?
  2. we wrote a python code similar with yours, the difference is using “fakesink” instead of “nvsmsgconv + nvmsgbroker”, but the app still crashed. here is the code. (30.2 KB)
  3. we wrote a C code similar with yours, the difference is using “fakesink” instead of “nvsmsgconv + nvmsgbroker”, but the app did not crash. here is the code and logs.
    deepstream_test3_app.c (33.8 KB)
    c_log.txt (49.5 KB)

To narrow down this issue, we need to check if sending fakesink is fine. from the test result, it seems be a a Python usage problem. will continue to check.

Thank you for investigating!
Could you clarify what you mean by “check if sending fakesink is fine?”

sorry I mean we can use fakesink to simplify the code.
As the point 2 said, from my side, if replacing “nvsmsgconv + nvmsgbroker” with fakesink in Python code, the Python app still crashed. so we can rule out the nvsmsgconv and nvmsgbroker’s issue. did you get the same result?

I’m not able to reproduce a seg fault with the app you provided after about 10 runs. I will keep trying, it may just be coincidence.

I’m running it by doing:

python3 --input file:///opt/nvidia/deepstream/deepstream-6.3/samples/streams/sample_1080p_h264

After multiple tries, I wasn’t able to reproduce it with the script you provided. I was able to reproduce it with my own script on the first run, however. It could be complete coincidence that it’s not showing up with your script, though.

Out of curiosity, have you tried remaking the app with msgconv and msgbroker in C? Does the problem show up in that case?

the correct reproducing command-line is
python3 --input file:///opt/nvidia/deepstream/deepstream-6.3/samples/streams/sample_1080p_h264.mp4 file:///opt/nvidia/deepstream/deepstream/samples/streams/sample_1080p_h265.mp4

Sorry, you’re right. I copy and pasted wrong, but that’s what I was running. I re-ran probably 20 more times this morning to be sure and still couldn’t cause the crash with the script you provided.

If the backtrace is the same, though, I’ll leave it up to you to decide what’s best. It may be that I’ve gotten “unlucky” and just not seen the seg fault by chance.

  1. Here is easier way to reproduce. please do the following modification. which is also the same with your code logic.
    if is_first_object and (frame_number % 2) == 0:
    if (frame_number % 2) == 0:
    please have a try and share the log. Thanks!
  2. if you can reproduce, we will continue to narrow down the issue.

Yep! Perfect! Reproduced on the first try. The backtrace looks the same as well.

1 Like

we have fixed this bug internally.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.