Information as applicable to my setup.
• Hardware Platform (Jetson / GPU) Jetson Nano B01
• DeepStream Version: 6.0.1
• JetPack Version (valid for Jetson only): 4.6.6
Hi, I use a USB camera Hikvision: so I use v4l2src with input type: jpeg 25fps. I try using decode by: jpegdec and nvjpegdec. I don’t know why ouput fps with nvjpegdec < jpegdec. Can you give me information or documents about it? (with nvjpegdec I add a nvvideoconvert after it)
This is my component with my USB Camera:
There is only jpegdec pipeline. What is the nvjpegdec pipeline?
What is the sink of your pipeline? How did you measure the output FPS?
Sorry, this is detail infor:
Pipeline:
jpegdec
Duration: 00:00:14.04, start: 0.000000, bitrate: 4066 kb/s
Stream #0:0(und): Video: h264 (Constrained Baseline) (avc1 / 0x31637661), yuv420p(bt709, progressive), 1280x720, 4063 kb/s, SAR 1:1 DAR 16:9, 25 fps, 250 tbr, 2500 tbn, 5k tbc (default)
nvjpegdec
Test by ffmpeg: 4.19 fps
Duration: 00:00:12.64, start: 0.000000, bitrate: 726 kb/s
Stream #0:0(und): Video: h264 (Constrained Baseline) (avc1 / 0x31637661), yuv420p(bt709, progressive), 1280x720, 725 kb/s, SAR 1:1 DAR 16:9, 4.19 fps, 250 tbr, 2500 tbn, 5k tbc (default)
This is my jpegdec code (comment videoconvert if testing nvjpegdec)
#include <gst/gst.h>
#include <signal.h>
static GstElement *pipeline = NULL;
void handle_sigint(int sig)
{
if (pipeline)
{
g_print("Caught SIGINT, sending EOS...\n");
gst_element_send_event(pipeline, gst_event_new_eos());
}
}
int main(int argc, char *argv[])
{
GstElement *source, *capsfilter, *decoder,
*videoconvert, *nvvidconv,
*encoder, *parser, *muxer, *sink;
GstCaps *caps;
GstBus *bus;
GstMessage *msg;
GstStateChangeReturn ret;
// Initialize GStreamer
gst_init(&argc, &argv);
// Set up the SIGINT handler
signal(SIGINT, handle_sigint);
// Create the elements
source = gst_element_factory_make("v4l2src", "source");
capsfilter = gst_element_factory_make("capsfilter", "capsfilter");
decoder = gst_element_factory_make("jpegdec", "decoder");
videoconvert = gst_element_factory_make("videoconvert", "videoconvert");
nvvidconv = gst_element_factory_make("nvvidconv", "nvvidconv");
encoder = gst_element_factory_make("nvv4l2h264enc", "encoder");
parser = gst_element_factory_make("h264parse", "parser");
muxer = gst_element_factory_make("qtmux", "muxer");
sink = gst_element_factory_make("filesink", "sink");
// Check if all elements are created
if (!source || !capsfilter || !decoder ||
!videoconvert ||
!nvvidconv ||
!encoder || !parser || !muxer || !sink)
{
g_printerr("Not all elements could be created.\n");
return -1;
}
// Create the empty pipeline
pipeline = gst_pipeline_new("video-pipeline");
if (!pipeline)
{
g_printerr("Pipeline could not be created.\n");
return -1;
}
// Set the element properties
g_object_set(source, "device", "/dev/video0", NULL);
caps = gst_caps_new_simple("image/jpeg",
"width", G_TYPE_INT, 1280,
"height", G_TYPE_INT, 720,
"framerate", GST_TYPE_FRACTION, 25, 1,
NULL);
g_object_set(capsfilter, "caps", caps, NULL);
gst_caps_unref(caps);
g_object_set(sink, "location", "output.mp4", NULL);
// Build the pipeline
gst_bin_add_many(GST_BIN(pipeline), source, capsfilter, decoder,
nvvidconv,
videoconvert,
encoder, parser, muxer, sink, NULL);
if (gst_element_link_many(source, capsfilter, decoder,
videoconvert,
nvvidconv,
encoder, parser, muxer, sink, NULL) != TRUE)
{
g_printerr("Elements could not be linked.\n");
gst_object_unref(pipeline);
return -1;
}
// Start playing
ret = gst_element_set_state(pipeline, GST_STATE_PLAYING);
if (ret == GST_STATE_CHANGE_FAILURE)
{
g_printerr("Unable to set the pipeline to the playing state.\n");
gst_object_unref(pipeline);
return -1;
}
// Wait until error or EOS
bus = gst_element_get_bus(pipeline);
msg = gst_bus_timed_pop_filtered(bus, GST_CLOCK_TIME_NONE, static_cast<GstMessageType>(GST_MESSAGE_ERROR | GST_MESSAGE_EOS));
// Parse message
if (msg != NULL)
{
GError *err;
gchar *debug_info;
switch (GST_MESSAGE_TYPE(msg))
{
case GST_MESSAGE_ERROR:
gst_message_parse_error(msg, &err, &debug_info);
g_printerr("Error received from element %s: %s\n", GST_OBJECT_NAME(msg->src), err->message);
g_printerr("Debugging information: %s\n", debug_info ? debug_info : "none");
g_clear_error(&err);
g_free(debug_info);
break;
case GST_MESSAGE_EOS:
g_print("End-Of-Stream reached.\n");
break;
default:
// We should not reach here because we only asked for ERRORs and EOS
g_printerr("Unexpected message received.\n");
break;
}
gst_message_unref(msg);
}
GST_DEBUG_BIN_TO_DOT_FILE(
GST_BIN(pipeline), GST_DEBUG_GRAPH_SHOW_ALL, "pipeline");
// Free resources
gst_object_unref(bus);
gst_element_set_state(pipeline, GST_STATE_NULL);
gst_object_unref(pipeline);
return 0;
}
I’ve tried with JetPack 6.1 and DeepStream 7.1 on Orin NX.
The nvjpegdec is faster than jpegdec with local video file.
The pipelines are
GST_DEBUG=fpsdisplaysink:7 gst-launch-1.0 filesrc location=/opt/nvidia/deepstream/deepstream/samples/streams/sample_720p_mjpeg.mp4 ! qtdemux name=d d.video_0 ! nvjpegdec ! fpsdisplaysink text-overlay=false video-sink=fakesink signal-fps-measurements=true sync=false
and
GST_DEBUG=fpsdisplaysink:7 gst-launch-1.0 filesrc location=/opt/nvidia/deepstream/deepstream/samples/streams/sample_720p_mjpeg.mp4 ! qtdemux name=d d.video_0 ! jpegdec ! fpsdisplaysink text-overlay=false video-sink=fakesink signal-fps-measurements=true sync=false
Seems the nvjpegdec is slower than jpegdec only with live stream. We will investigate it.
Tks! If you have more infor about it. Please let me know =))
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.