GstInference/Benchmarks: Difference between revisions

From RidgeRun Developer Wiki
Line 1,139: Line 1,139:
== TensorRT Benchmarks ==
== TensorRT Benchmarks ==


=== FPS Measurements ===


<html>
<style>
    .button {
    background-color: #008CBA;
    border: none;
    color: white;
    padding: 15px 32px;
    text-align: center;
    text-decoration: none;
    display: inline-block;
    font-size: 16px;
    margin: 4px 2px;
    cursor: pointer;
  }
</style>
<div id="Buttons_Model_Tensorrt_Fps" style="margin: auto; width: 1300px; height: auto;">
<button class="button" id="show_inceptionv1_tensorrt_fps">Show InceptionV1 </button>
<button class="button" id="show_inceptionv2_tensorrt_fps">Show InceptionV2 </button>
<button class="button" id="show_inceptionv3_tensorrt_fps">Show InceptionV3 </button>
<button class="button" id="show_inceptionv4_tensorrt_fps">Show InceptionV4 </button>
<button class="button" id="show_tinyyolov2_tensorrt_fps">Show TinyYoloV2 </button>
<button class="button" id="show_tinyyolov3_tensorrt_fps">Show TinyYoloV3 </button>
</div>
<br><br>
<div id="chart_tensorrtrt" style="margin: auto; width: 800px; height: 500px;"></div>
<br><br>
<div id="Buttons_Backend_TensorRT_Fps" style="margin: auto; width: 600px; height: auto;">
<button class="button" id="show_tensorrt_fps">Show TensorRT </button>
</div>
<div id="chart_tensorrt1" style="margin: auto; width: 800px; height: 500px;"></div>
<br><br>
<script>
      google.charts.load('current', {'packages':['corechart', 'bar']});
      google.charts.setOnLoadCallback(drawStuffFpsTensorRT);
     
      function drawStuffFpsTensorRT() {
        var tensorrt_chartDiv = document.getElementById('chart_tensorrt');
        var tensorrt_chartDiv1 = document.getElementById('chart_tensorrt1');
        var table_backend_platform_tensorrt_fps = google.visualization.arrayToDataTable([
          ['Platform',                  //Column 0
          'InceptionV1 \n TensorRT',    //Column 1
          'InceptionV2 \n TensorRT',    //Column 2
          'InceptionV3 \n TensorRT',    //Column 3
          'InceptionV4 \n TensorRT',    //Column 4
          'TinyYoloV2 \n TensorRT',      //Column 5
          'TinyYoloV3 \n TensorRT'],    //Column 6
          ['x86', 47.8702, 32.7236, 12.092, 5.2632, 16.03, 18.3592]
        ]);
        var table_model_platform_tensorrt_fps = google.visualization.arrayToDataTable([
          ['Model',                  //Column 0
            'TensorRT \n x86'],        //Column 1
          ['InceptionV1', 47.8702], //row 1
          ['InceptionV2', 32.7236], //row 2
          ['InceptionV3', 12.092], //row 3
          ['InceptionV4', 5.2632], //row 4
          ['TinyYoloV2',  16.03], //row 5
          ['TinyYoloV3',  18.3592]  //row 6
        ]);
        var tensorrt_materialOptions = {
          width: 320,
          chart: {
            title: 'Backend Vs Platform per model',
          },
          series: {
          },
          axes: {
            y: {
              distance: {side: 'left',label: 'FPS'}, // Left y-axis.
            }
          }
        };
        var tensorrt_materialOptions1 = {
          width: 900,
          chart: {
            title: 'Model Vs Platform per backend',
          },
          series: {
          },
          axes: {
            y: {
              distance: {side: 'left',label: 'FPS'}, // Left y-axis.
            }
          }
        };
        var materialChart_tensorrt_fps = new google.charts.Bar(tensorrt_chartDiv);
        var materialChart_tensorrt_fps1 = new google.charts.Bar(tensorrt_chartDiv1);
        view_tensorrt_fps = new google.visualization.DataView(table_backend_platform_tensorrt_fps);
        view_tensorrt_fps1 = new google.visualization.DataView(table_model_platform_tensorrt_fps);
        function drawMaterialChart() {
          var materialChart_tensorrt_fps = new google.charts.Bar(tensorrt_chartDiv);
          var materialChart_tensorrt_fps1 = new google.charts.Bar(tensorrt_chartDiv1);
          materialChart_tensorrt_fps.draw(table_backend_platform_tensorrt_fps, google.charts.Bar.convertOptions(tensorrt_materialOptions));
          materialChart_tensorrt_fps1.draw(table_model_platform_tensorrt_fps, google.charts.Bar.convertOptions(tensorrt_materialOptions1));
          init_charts();
        }
        function init_charts(){
          view_tensorrt_fps.setColumns([0,1]);
          view_tensorrt_fps.hideColumns([2, 3, 4, 5, 6]);
          materialChart_tensorrt_fps.draw(view_tensorrt_fps, tensorrt_materialOptions);
          view_tensorrt_fps1.setColumns([0,1]);
          materialChart_tensorrt_fps1.draw(view_tensorrt_fps1, tensorrt_materialOptions1);
        }
        // REF_MODEL
        /*Select the Model that you want to show in the chart*/
        var show_inceptionv1_tensorrt_fps = document.getElementById('show_inceptionv1_tensorrt_fps');
        show_inceptionv1_tensorrt_fps.onclick = function () {
          view_tensorrt_fps.setColumns([0,1]);
          view_tensorrt_fps.hideColumns([2, 3, 4, 5, 6]);
          materialChart_tensorrt_fps.draw(view_tensorrt_fps, tensorrt_materialOptions);
        }
        var show_inceptionv2_tensorrt_fps = document.getElementById('show_inceptionv2_tensorrt_fps');
        show_inceptionv2_tensorrt_fps.onclick = function () {
          view_tensorrt_fps.setColumns([0,2]);
          view_tensorrt_fps.hideColumns([1, 3, 4, 5, 6]);
          materialChart_tensorrt_fps.draw(view_tensorrt_fps, tensorrt_materialOptions);
        }
        var show_inceptionv3_tensorrt_fps = document.getElementById('show_inceptionv3_tensorrt_fps');
        show_inceptionv3_tensorrt_fps.onclick = function () {
          view_tensorrt_fps.setColumns([0,3]);
          view_tensorrt_fps.hideColumns([1, 2, 4, 5, 6]);
          materialChart_tensorrt_fps.draw(view_tensorrt_fps, tensorrt_materialOptions);
        }
        var show_inceptionv4_tensorrt_fps = document.getElementById('show_inceptionv4_tensorrt_fps');
        show_inceptionv4_tensorrt_fps.onclick = function () {
          view_tensorrt_fps.setColumns([0,4]);
          view_tensorrt_fps.hideColumns([1, 2, 3, 5, 6]);
          materialChart_tensorrt_fps.draw(view_tensorrt_fps, tensorrt_materialOptions);
        }
        var show_tinyyolov2_tensorrt_fps = document.getElementById('show_tinyyolov2_tensorrt_fps');
        show_tinyyolov2_tensorrt_fps.onclick = function () {
          view_tensorrt_fps.setColumns([0,5]);
          view_tensorrt_fps.hideColumns([1, 2, 3, 4, 6]);
          materialChart_tensorrt_fps.draw(view_tensorrt_fps, tensorrt_materialOptions);
        }
        var show_tinyyolov3_tensorrt_fps = document.getElementById('show_tinyyolov3_tensorrt_fps');
        show_tinyyolov3_tensorrt_fps.onclick = function () {
          view_tensorrt_fps.setColumns([0,6]);
          view_tensorrt_fps.hideColumns([1, 2, 3, 4, 5]);
          materialChart_tensorrt_fps.draw(view_tensorrt_fps, tensorrt_materialOptions);
        }
        // REF_BACKEND
        /*Select the Model that you want to show in the chart*/
        var show_tensorrt_fps = document.getElementById('show_tensorrt_fps');
        show_tensorrt_fps.onclick = function () {
          view_tensorrt_fps1.setColumns([0,1]);
          materialChart_tensorrt_fps1.draw(view_tensorrt_fps1, tensorrt_materialOptions1);
        }
        drawMaterialChart();
    };
</script>
</html>




Revision as of 23:48, 8 July 2020




Previous: Example Applications/DispTec Index Next: Model Zoo


GstInference
GstInference
Introduction
Getting started
Supported architectures

InceptionV1

InceptionV3

YoloV2

AlexNet
Supported backends

Caffe

Metadata and Signals
Overlay Elements
Utils Elements
Legacy pipelines
Example pipelines
Example applications
Benchmarks
Model Zoo
Project Status
Contact Us

Coral from Google




GstInference Benchmarks

The following benchmarks were run with a source video (1920x1080@60). With the following base GStreamer pipeline, and environment variables: 

$ VIDEO_FILE='video.mp4'
$ MODEL_LOCATION='graph_inceptionv1_tensorflow.pb'
$ INPUT_LAYER='input'
$ OUTPUT_LAYER='InceptionV1/Logits/Predictions/Reshape_1'

The environment variables were changed accordingly with the used model (Inception V1,V2,V3 or V4) 

GST_DEBUG=inception1:1 gst-launch-1.0 filesrc location=$VIDEO_FILE ! decodebin ! videoconvert ! videoscale ! queue ! net.sink_model inceptionv1 name=net model-location=$MODEL_LOCATION backend=tensorflow backend::input-layer=$INPUT_LAYER  backend::output-layer=$OUTPUT_LAYER net.src_model ! perf ! fakesink -v

The Desktop PC had the following specifications:

  • Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz
  • 8 GB RAM
  • Cedar [Radeon HD 5000/6000/7350/8350 Series]
  • Linux 4.15.0-54-generic x86_64 (Ubuntu 16.04)

The Jetson Xavier power modes used were 2 and 6 (more information: Supported Modes and Power Efficiency)

  • View current power mode:
$ sudo /usr/sbin/nvpmodel -q
  • Change current power mode:
sudo /usr/sbin/nvpmodel -m x

Where x is the power mode ID (e.g. 0, 1, 2, 3, 4, 5, 6).

Summary

Desktop PC CPU Library
Model Framerate CPU Usage
Inception V1 11.89 48
Inception V2 10.33 65
Inception V3 5.41 90
Inception V4 3.81 94
Jetson Xavier (15W) CPU Library GPU Library
Model Framerate CPU Usage Framerate CPU Usage
Inception V1 8.24 86 52.3 43
Inception V2 6.58 88 39.6 42
Inception V3 2.54 92 17.8 25
Inception V4 1.22 94 9.4 20
Jetson Xavier (30W) CPU Library GPU Library
Model Framerate CPU Usage Framerate CPU Usage
Inception V1 6.41 93 66.27 72
Inception V2 5.11 95 50.59 62
Inception V3 1.96 98 22.95 44
Inception V4 0.98 99 12.14 32

Framerate

thumb
thumb

CPU Usage

thumb
thumb

TensorFlow Lite Benchmarks

FPS measurement







CPU usage measurement







Test benchmark video

The following video was used to perform the benchmark tests.
To download the video press right click on the video and select 'Save video as' and save this in your computer.

Test benchmark video

ONNXRT Benchmarks

The Desktop PC had the following specifications:

  • Intel(R) Core(TM) Core i7-7700HQ CPU @ 2.80GHz
  • 12 GB RAM
  • Linux 4.15.0-106-generic x86_64 (Ubuntu 16.04)
  • GStreamer 1.8.3

The following was the GStreamer pipeline used to obtain the results: 

# MODELS_PATH has the following structure
#/path/to/models/
#├── InceptionV1_onnxrt
#│   ├── graph_inceptionv1_info.txt
#│   ├── graph_inceptionv1.onnx
#│   └── labels.txt
#├── InceptionV2_onnxrt
#│   ├── graph_inceptionv2_info.txt
#│   ├── graph_inceptionv2.onnx
#│   └── labels.txt
#├── InceptionV3_onnxrt
#│   ├── graph_inceptionv3_info.txt
#│   ├── graph_inceptionv3.onnx
#│   └── labels.txt
#├── InceptionV4_onnxrt
#│   ├── graph_inceptionv4_info.txt
#│   ├── graph_inceptionv4.onnx
#│   └── labels.txt
#├── TinyYoloV2_onnxrt
#│   ├── graph_tinyyolov2_info.txt
#│   ├── graph_tinyyolov2.onnx
#│   └── labels.txt
#└── TinyYoloV3_onnxrt
#    ├── graph_tinyyolov3_info.txt
#    ├── graph_tinyyolov3.onnx
#    └── labels.txt

model_array=(inceptionv1 inceptionv2 inceptionv3 inceptionv4 tinyyolov2 tinyyolov3)
model_upper_array=(InceptionV1 InceptionV2 InceptionV3 InceptionV4 TinyYoloV2 TinyYoloV3)
MODELS_PATH=/path/to/models/
INTERNAL_PATH=onnxrt
EXTENSION=".onnx"

gst-launch-1.0 \
filesrc location=$VIDEO_PATH num-buffers=600 ! decodebin ! videoconvert ! \
perf print-arm-load=true name=inputperf ! tee name=t t. ! videoscale ! queue ! net.sink_model t. ! queue ! net.sink_bypass \
${model_array[i]} backend=onnxrt name=net \
model-location="${MODELS_PATH}${model_upper_array[i]}_${INTERNAL_PATH}/graph_${model_array[i]}${EXTENSION}" \
net.src_bypass ! perf print-arm-load=true name=outputperf ! videoconvert ! fakesink sync=false

FPS Measurements







CPU Load Measurements







Test benchmark video

The following video was used to perform the benchmark tests.
To download the video press right click on the video and select 'Save video as' and save this in your computer.

Test benchmark video

TensorRT Benchmarks

Previous: Example Applications/DispTec Index Next: Model Zoo



Retrieved from "https://developer.ridgerun.com/wiki/index.php?title=GstInference/Benchmarks&oldid=31256"