Quick Start Guide for Media Analytics on Intel® Data Center GPU Flex Series

Contents

Quick Start Guide for Media Analytics on Intel® Data Center GPU Flex Series#

These steps provide instructions on how to download and launch the Media Analytics container and run an object detection pipeline using Intel® Deep Learning Streamer (Intel® DL Streamer) on Intel® Data Center GPU Flex Series.

Optimized For	Description
OS	Ubuntu* 22.04
What You Will Learn	How to launch a basic object detection pipeline.
Time to Complete	10 minutes

Purpose#

pipeline diagram

This quick start guide uses the gst-launch-1.0 utility to launch a simple media analytics pipeline performing:

Media decode
Inference using MobileNetV2 + SSD-based network for Person/Vehicle/Bike Detection
Overlay of detected objects.

Key Implementation Details#

Configuration	Default Setting
Target device	Intel® Data Center GPU Flex Series
Input format	mp4
Output format	XDisplay or mp4
Output resolution	same as input

Perform the following steps on a Linux Ubuntu 22.04 System#

Step 1: Allow connection to X server#

xhost local:root
setfacl -m user:1000:r ~/.Xauthority

Step 2: Launch Intel® DL Streamer Container#

DEVICE=${DEVICE:-/dev/dri/renderD128}

DEVICE_GRP=$(ls -g $DEVICE | awk '{print $3}' | \
  xargs getent group | awk -F: '{print $3}')

docker run -it --rm --net=host -e no_proxy=$no_proxy -e https_proxy=$https_proxy -e socks_proxy=$socks_proxy -e http_proxy=$http_proxy \
-v ~/.Xauthority:/home/dlstreamer/.Xauthority -v /tmp/.X11-unix -e DISPLAY=$DISPLAY \
--device $DEVICE --group-add $DEVICE_GRP \
intel/dlstreamer:dgpu-dpcpp-devel /bin/bash

Step 3: Download Sample Media#

wget "https://www.pexels.com/video/1721294/download/?w=640&h=360" -O pexels_1721294.mp4

Step 4: Download Object Detection Model#

omz_downloader --name person-vehicle-bike-detection-crossroad-1016

Step 5(a): Run Object Detection Pipeline and Display Results#

gst-launch-1.0 filesrc location=pexels_1721294.mp4 ! decodebin ! video/x-raw\(memory:VASurface\) ! \
               gvadetect model=./intel/person-vehicle-bike-detection-crossroad-1016/FP16-INT8/person-vehicle-bike-detection-crossroad-1016.xml \
                         model-proc=/opt/intel/dlstreamer/samples/gstreamer/model_proc/intel/person-vehicle-bike-detection-crossroad-1016.json \
                         pre-process-backend=vaapi-surface-sharing \
                         device=GPU ! \
                         meta_overlay device=GPU ! \
               videoconvert ! ximagesink

Step 5(b): Run Object Detection Pipeline and Save Results#

gst-launch-1.0 filesrc location=pexels_1721294.mp4 ! decodebin ! video/x-raw\(memory:VASurface\) ! \
               gvadetect model=./intel/person-vehicle-bike-detection-crossroad-1016/FP16-INT8/person-vehicle-bike-detection-crossroad-1016.xml \
                         model-proc=/opt/intel/dlstreamer/samples/gstreamer/model_proc/intel/person-vehicle-bike-detection-crossroad-1016.json \
                         pre-process-backend=vaapi-surface-sharing \
                         device=GPU ! \
               meta_overlay device=GPU ! \
               gvametaconvert json-indent=4 ! gvametapublish file-path=detection_results.json ! \
               vaapih264enc ! h264parse ! mp4mux ! filesink location=detection_output.mp4

*Other names and brands may be claimed as the property of others.