20,290
edits
DeepStream 6.3 pipelines: Difference between revisions
m
no edit summary
mNo edit summary |
|||
| Line 2: | Line 2: | ||
<div style="float:right; clear:both; margin-left:0.5em;">__TOC__</div> | <div style="float:right; clear:both; margin-left:0.5em;">__TOC__</div> | ||
==Introduction== | ==Introduction== | ||
On this page, you are going to find a set of DeepStream pipelines used on Jetson Xavier NX DevKit. This wiki page tries to describe some of the DeepStream features for the NVIDIA platforms and other multimedia features. The goal is to provide you some example pipelines. A general NVIDIA Deepstream overview can be found at: [https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_Overview.html NVIDIA Deepstream overview]. | On this page, you are going to find a set of DeepStream pipelines used on Jetson Xavier NX DevKit. This wiki page tries to describe some of the DeepStream features for the NVIDIA platforms and other multimedia features. The goal is to provide you some example pipelines. A general NVIDIA Deepstream overview can be found at: [https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_Overview.html NVIDIA Deepstream overview]. | ||
| Line 85: | Line 86: | ||
* '''Using RtspSink''': | * '''Using RtspSink''': | ||
For convenience, the following pipeline is provided in case you want to check the compatibility of deepstream with [ | For convenience, the following pipeline is provided in case you want to check the compatibility of deepstream with [[GstRtspSink | rtspsink]]. | ||
gst-launch-1.0 filesrc location=/opt/nvidia/deepstream/deepstream/samples/streams/sample_720p.mp4 ! qtdemux ! h264parse ! nvv4l2decoder ! m.sink_0 nvstreammux name=m batch-size=1 width=1920 height=1080 ! nvinfer config-file-path=/opt/nvidia/deepstream/deepstream/sources/apps/sample_apps/deepstream-test1/dstest1_pgie_config.yml ! nvtracker ll-lib-file=/opt/nvidia/deepstream/deepstream/lib/libnvds_nvmultiobjecttracker.so ! nvvideoconvert ! nvdsosd ! nvvideoconvert ! nvv4l2h264enc insert-sps-pps=true idrinterval=30 ! video/x-h264, mapping=/stream1 ! perf ! rtspsink service=49152 | gst-launch-1.0 filesrc location=/opt/nvidia/deepstream/deepstream/samples/streams/sample_720p.mp4 ! qtdemux ! h264parse ! nvv4l2decoder ! m.sink_0 nvstreammux name=m batch-size=1 width=1920 height=1080 ! nvinfer config-file-path=/opt/nvidia/deepstream/deepstream/sources/apps/sample_apps/deepstream-test1/dstest1_pgie_config.yml ! nvtracker ll-lib-file=/opt/nvidia/deepstream/deepstream/lib/libnvds_nvmultiobjecttracker.so ! nvvideoconvert ! nvdsosd ! nvvideoconvert ! nvv4l2h264enc insert-sps-pps=true idrinterval=30 ! video/x-h264, mapping=/stream1 ! perf ! rtspsink service=49152 | ||
| Line 117: | Line 118: | ||
====Primary_Detector + Overlay + tracker (libnvds_nvmultiobjecttracker) + Secondary_VehicleTypes + Secondary_CarColor with GstD==== | ====Primary_Detector + Overlay + tracker (libnvds_nvmultiobjecttracker) + Secondary_VehicleTypes + Secondary_CarColor with GstD==== | ||
Alternatively, you can use three types of inference in a single pipeline. This time, the solution is presented using GstD so you can see its use with Deepstream (GstD is a more flexible and powerful version of gst-launch; you can read more about it here: [ | Alternatively, you can use three types of inference in a single pipeline. This time, the solution is presented using GstD so you can see its use with Deepstream (GstD is a more flexible and powerful version of gst-launch; you can read more about it here: [[GStreamer%20Daemon | gstd]]). As you can see in the resulting video, the pipeline identifies three things: | ||
* It is a car. | * It is a car. | ||
* It is a truck. | * It is a truck. | ||