Update to 1.1.2 release

This release is compatible with DeepStream SDK 6.1

Ubuntu 20.04
Python 3.8
DeepStream SDK 6.1

Features:
- New Preprocess test app which demonstrates using nvdspreprocess plugin with custom ROIs
- Enhanced Test3 to support Triton, no-display mode, file-loop, and silent mode
- Minor improvements and bug fixes

HOWTO.md

@@ -15,10 +15,10 @@ This guide provides resources for DeepStream application development in Python.
 <a name="prereqs"></a>
 ## Prerequisites
 
-* Ubuntu 18.04
-* [DeepStream SDK 6.0.1](https://developer.nvidia.com/deepstream-download) or later
-* Python 3.6+
-* [Gst Python](https://gstreamer.freedesktop.org/modules/gst-python.html) v1.14.5
+* Ubuntu 20.04
+* [DeepStream SDK 6.1](https://developer.nvidia.com/deepstream-download) or later
+* Python 3.8
+* [Gst Python](https://gstreamer.freedesktop.org/modules/gst-python.html) v1.16.2
 
 Gst python should be already installed on Jetson.  
 If missing, install with the following steps:
@@ -28,7 +28,7 @@ If missing, install with the following steps:
    $ export GST_CFLAGS="-pthread -I/usr/include/gstreamer-1.0 -I/usr/include/glib-2.0 -I/usr/lib/x86_64-linux-gnu/glib-2.0/include"
    $ git clone https://github.com/GStreamer/gst-python.git
    $ cd gst-python
-   $ git checkout 1a8f48a
+   $ git checkout 5343aeb
    $ ./autogen.sh PYTHON=python3
    $ ./configure PYTHON=python3
    $ make
@@ -45,11 +45,11 @@ Note: Compiling bindings now also generates a pip installable python wheel for t
 <a name="run_samples"></a>
 ## Running Sample Applications
 
-Clone the deepstream_python_apps repo under <DeepStream 6.0.1 ROOT>/sources:
+Clone the deepstream_python_apps repo under <DeepStream 6.1 ROOT>/sources:
 git clone https://github.com/NVIDIA-AI-IOT/deepstream_python_apps
 
 This will create the following directory:  
-```<DeepStream 6.0.1 ROOT>/sources/deepstream_python_apps```  
+```<DeepStream 6.1 ROOT>/sources/deepstream_python_apps```
 
 The Python apps are under the "apps" directory.  
 Go into each app directory and follow instructions in the README.  

README.md

@@ -2,9 +2,11 @@
 
 This repository contains Python bindings and sample applications for the [DeepStream SDK](https://developer.nvidia.com/deepstream-sdk).  
 
-SDK version supported: 6.0.1
+SDK version supported: 6.1
 
-<b>NEW: The bindings sources along with build instructions are now available under [bindings](bindings)!  </b>
+<b>The bindings sources along with build instructions are now available under [bindings](bindings)!  </b>
+
+<b>This release comes with Operating System upgrades (from Ubuntu 18.04 to Ubuntu 20.04) for DeepStreamSDK 6.1 support. This translates to upgrade in Python version to 3.8 and [gst-python](3rdparty/gst-python/) version has also been upgraded to 1.16.2 !</b>
 
 Download the latest release package complete with bindings and sample applications from the [release section](../../releases).  
 
@@ -41,7 +43,7 @@ To run the sample applications or write your own, please consult the [HOW-TO Gui
 We currently provide the following sample applications:
 * [deepstream-test1](apps/deepstream-test1) -- 4-class object detection pipeline
 * [deepstream-test2](apps/deepstream-test2) -- 4-class object detection, tracking and attribute classification pipeline
-* [deepstream-test3](apps/deepstream-test3) -- multi-stream pipeline performing 4-class object detection
+* <b>UPDATE</b> [deepstream-test3](apps/deepstream-test3) -- multi-stream pipeline performing 4-class object detection - now also supports triton inference server, no-display mode, file-loop and silent mode
 * [deepstream-test4](apps/deepstream-test4) -- msgbroker for sending analytics results to the cloud
 * [deepstream-imagedata-multistream](apps/deepstream-imagedata-multistream) -- multi-stream pipeline with access to image buffers
 * [deepstream-ssd-parser](apps/deepstream-ssd-parser) -- SSD model inference via Triton server with output parsing in Python
@@ -53,6 +55,7 @@ We currently provide the following sample applications:
 * [runtime_source_add_delete](apps/runtime_source_add_delete) -- add/delete source streams at runtime
 * [deepstream-imagedata-multistream-redaction](apps/deepstream-imagedata-multistream-redaction) -- multi-stream pipeline with face detection and redaction
 * [deepstream-rtsp-in-rtsp-out](apps/deepstream-rtsp-in-rtsp-out) -- multi-stream pipeline with RTSP input/output
+* <b>NEW</b> [deepstream-preprocess-test](apps/deepstream-preprocess-test) -- multi-stream pipeline using nvdspreprocess plugin with custom ROIs
 
 
 Detailed application information is provided in each application's subdirectory under [apps](apps).  

apps/README

@@ -19,52 +19,46 @@
 DeepStream SDK Python Bindings
 ================================================================================
 Setup pre-requisites:
-- Ubuntu 18.04
-- NVIDIA DeepStream SDK 6.0.1
-- Python 3.6
+- Ubuntu 20.04
+- NVIDIA DeepStream SDK 6.1
+- Python 3.8
 - Gst-python
 
 --------------------------------------------------------------------------------
 Package Contents
 --------------------------------------------------------------------------------
-The DeepStream Python package includes:
-1. Python bindings for DeepStream Metadata libraries
-   These bindings are installed as part of the SDK at:
-   /opt/nvidia/deepstream/deepstream/lib/pyds.so
-
-   Sample applications that import is_aarch_64 automatically
-   have this path added.
+1. DeepStream Python bindings located in bindings dir
+   with installation instructions in bindings/README.md
 
-   A setup.py is also provided to install this extension into standard path.
-   Currently this needs to be run manually:
-   $ cd /opt/nvidia/deepstream/deepstream/lib
-   $ python3 setup.py install
-
 2. DeepStream test apps in Python
    The following test apps are available:
    deepstream-test1
    deepstream-test2
    deepstream-test3
    deepstream-test4
    deepstream-imagedata-multistream
+   deepstream-imagedata-multistream-redaction
    deepstream-ssd-parser
    deepstream-test1-rtsp-out
+   deepstream-rtsp-in-rtsp-out
    deepstream-test1-usbcam
    deepstream-opticalflow
    deepstream-segmentation
    deepstream-nvdsanalytics
+   deepstream-preprocess-test
+   runtime_source_add_delete
 
 --------------------------------------------------------------------------------
 Installing Pre-requisites:
 --------------------------------------------------------------------------------
 
-DeepStream SDK 6.0.1
+DeepStream SDK 6.1
 --------------------
 Download and install from https://developer.nvidia.com/deepstream-download
 
-Python 3.6
+Python 3.8
 ----------
-Should be already installed with Ubuntu 18.04
+Should be already installed with Ubuntu 20.04
 
 Gst-python
 ----------
@@ -76,7 +70,7 @@ $ sudo apt install python3-gi python3-dev python3-gst-1.0 -y
 --------------------------------------------------------------------------------
 Running the samples
 --------------------------------------------------------------------------------
-The apps are configured to work from inside the DeepStream SDK 6.0.1 installation.
+The apps are configured to work from inside the DeepStream SDK 6.1 installation.
 
 Clone the deepstream_python_apps repo under <DeepStream ROOT>/sources:
 $ git clone https://github.com/NVIDIA-AI-IOT/deepstream_python_apps

apps/common/FPS.py

@@ -16,30 +16,52 @@
 ################################################################################
 
 import time
+from threading import Lock
 start_time=time.time()
-frame_count=0
+
+fps_mutex = Lock()
 
 class GETFPS:
     def __init__(self,stream_id):
         global start_time
         self.start_time=start_time
         self.is_first=True
-        global frame_count
-        self.frame_count=frame_count
+        self.frame_count=0
         self.stream_id=stream_id
-    def get_fps(self):
-        end_time=time.time()
-        if(self.is_first):
-            self.start_time=end_time
-            self.is_first=False
-        if(end_time-self.start_time>5):
-            print("**********************FPS*****************************************")
-            print("Fps of stream",self.stream_id,"is ", float(self.frame_count)/5.0)
-            self.frame_count=0
-            self.start_time=end_time
+
+    def update_fps(self):
+        end_time = time.time()
+        if self.is_first:
+            self.start_time = end_time
+            self.is_first = False
         else:
-            self.frame_count=self.frame_count+1
+            global fps_mutex
+            with fps_mutex:
+                self.frame_count = self.frame_count + 1
+
+    def get_fps(self):
+        end_time = time.time()
+        with fps_mutex:
+            stream_fps = float(self.frame_count/(end_time - self.start_time))
+            self.frame_count = 0
+        self.start_time = end_time
+        return round(stream_fps, 2)
+
     def print_data(self):
         print('frame_count=',self.frame_count)
         print('start_time=',self.start_time)
 
+class PERF_DATA:
+    def __init__(self, num_streams=1):
+        self.perf_dict = {}
+        self.all_stream_fps = {}
+        for i in range(num_streams):
+            self.all_stream_fps["stream{0}".format(i)]=GETFPS(i)
+
+    def perf_print_callback(self):
+        self.perf_dict = {stream_index:stream.get_fps() for (stream_index, stream) in self.all_stream_fps.items()}
+        print ("\n**PERF: ", self.perf_dict, "\n")
+        return True
+
+    def update_fps(self, stream_index):
+        self.all_stream_fps[stream_index].update_fps()

apps/common/bus_call.py

@@ -18,7 +18,7 @@
 import gi
 import sys
 gi.require_version('Gst', '1.0')
-from gi.repository import GObject, Gst
+from gi.repository import Gst
 def bus_call(bus, message, loop):
     t = message.type
     if t == Gst.MessageType.EOS:

apps/deepstream-imagedata-multistream-redaction/README

@@ -16,8 +16,8 @@
 ################################################################################
 
 Prerequisites:
-- DeepStreamSDK 6.0.1
-- Python 3.6
+- DeepStreamSDK 6.1
+- Python 3.8
 - Gst-python
 - NumPy package
 - OpenCV package
@@ -39,10 +39,8 @@ Yet, we need to install the introspection typelib package:
   $ sudo apt-get install gobject-introspection gir1.2-gst-rtsp-server-1.0
 
 Download Peoplenet model:
-  $ cd /opt/nvidia/deepstream/deepstream/samples/configs/tao_pretrained_models
-  $ mkdir -p ../../models/tao_pretrained_models/peoplenet && wget https://api.ngc.nvidia.com/v2/models/nvidia/tao/peoplenet/versions/pruned_v2.1/files/resnet34_peoplenet_pruned.etlt \
-    -O ../../models/tao_pretrained_models/peoplenet/resnet34_peoplenet_pruned.etlt
-  $ wget https://api.ngc.nvidia.com/v2/models/nvidia/tao/peoplenet/versions/pruned_v2.1/files/labels.txt -O ../../../configs/tao_pretrained_models/labels_peoplenet.txt
+  Please follow instructions from the README.md located at : /opt/nvidia/deepstream/deepstream/samples/configs/tao_pretrained_models/README.md
+  to download latest supported peoplenet model (V2.5 for this release)
 
 To run:
   $ python3 deepstream_imagedata-multistream_redaction.py -i <uri1> [uri2] ... [uriN] -c {H264,H265} -b BITRATE

apps/deepstream-imagedata-multistream-redaction/config_infer_primary_peoplenet.txt

@@ -19,9 +19,10 @@
 gpu-id=0
 net-scale-factor=0.0039215697906911373
 tlt-model-key=tlt_encode
-tlt-encoded-model=../../../../samples/models/tao_pretrained_models/peoplenet/resnet34_peoplenet_pruned.etlt
-labelfile-path=../../../../samples/configs/tao_pretrained_models/labels_peoplenet.txt
-model-engine-file=../../../../samples/models/tao_pretrained_models/peoplenet/resnet34_peoplenet_pruned.etlt_b1_gpu0_fp16.engine
+tlt-encoded-model=../../../../samples/models/tao_pretrained_models/peopleNet/V2.5/resnet34_peoplenet_int8.etlt
+labelfile-path=../../../../samples/models/tao_pretrained_models/peopleNet/V2.5/labels.txt
+model-engine-file=../../../../samples/models/tao_pretrained_models/peopleNet/V2.5/resnet34_peoplenet_int8.etlt_b1_gpu0_int8.engine
+int8-calib-file=../../../../samples/models/tao_pretrained_models/peopleNet/V2.5/resnet34_peoplenet_int8.txt
 input-dims=3;544;960;0
 uff-input-blob-name=input_1
 batch-size=1

apps/deepstream-imagedata-multistream-redaction/deepstream_imagedata-multistream_redaction.py

@@ -26,8 +26,7 @@
 
 gi.require_version('Gst', '1.0')
 gi.require_version('GstRtspServer', '1.0')
-from gi.repository import GObject, Gst, GstRtspServer
-from gi.repository import GLib
+from gi.repository import GLib, Gst, GstRtspServer
 from ctypes import *
 import time
 import sys
@@ -36,15 +35,15 @@
 from common.is_aarch_64 import is_aarch64
 from common.bus_call import bus_call
 
-from common.FPS import GETFPS
+from common.FPS import PERF_DATA
 import numpy as np
 import pyds
 import cv2
 import os
 import os.path
 from os import path
 
-fps_streams = {}
+perf_data = None
 frame_count = {}
 saved_count = {}
 global PGIE_CLASS_ID_PERSON
@@ -152,8 +151,10 @@ def tiler_sink_pad_buffer_probe(pad, info, u_data):
 
         print("Frame Number=", frame_number, "Number of Objects=", num_rects, "Face_count=",
               obj_counter[PGIE_CLASS_ID_FACE], "Person_count=", obj_counter[PGIE_CLASS_ID_PERSON])
-        # Get frame rate through this probe
-        fps_streams["stream{0}".format(frame_meta.pad_index)].get_fps()
+        # Update frame rate through this probe
+        stream_index = "stream{0}".format(frame_meta.pad_index)
+        global perf_data
+        perf_data.update_fps(stream_index)
         if save_image:
             img_path = "{}/stream_{}/frame_{}.jpg".format(folder_name, frame_meta.pad_index, frame_number)
             cv2.imwrite(img_path, frame_copy)
@@ -247,12 +248,11 @@ def create_source_bin(index, uri):
         return None
     return nbin
 
-
 def main(uri_inputs,codec,bitrate ):
     # Check input arguments
     number_sources = len(uri_inputs)
-    for i in range(0, number_sources ):
-        fps_streams["stream{0}".format(i)] = GETFPS(i)
+    global perf_data
+    perf_data = PERF_DATA(number_sources)
 
     global folder_name
     folder_name = "out_crops"
@@ -264,7 +264,6 @@ def main(uri_inputs,codec,bitrate ):
     os.mkdir(folder_name)
     print("Frames will be saved in ", folder_name)
     # Standard GStreamer initialization
-    GObject.threads_init()
     Gst.init(None)
 
     # Create gstreamer elements */
@@ -352,7 +351,7 @@ def main(uri_inputs,codec,bitrate ):
     if is_aarch64():
         encoder.set_property('preset-level', 1)
         encoder.set_property('insert-sps-pps', 1)
-        encoder.set_property('bufapi-version', 1)
+        #encoder.set_property('bufapi-version', 1)
 
     # Make the payload-encode video into RTP packets
     if codec == "H264":
@@ -430,7 +429,7 @@ def main(uri_inputs,codec,bitrate ):
     rtppay.link(sink)
 
     # create an event loop and feed gstreamer bus mesages to it
-    loop = GObject.MainLoop()
+    loop = GLib.MainLoop()
     bus = pipeline.get_bus()
     bus.add_signal_watch()
     bus.connect("message", bus_call, loop)
@@ -454,6 +453,8 @@ def main(uri_inputs,codec,bitrate ):
         sys.stderr.write(" Unable to get sink pad \n")
     else:
         tiler_sink_pad.add_probe(Gst.PadProbeType.BUFFER, tiler_sink_pad_buffer_probe, 0)
+        # perf callback function to print fps every 5 sec
+        GLib.timeout_add(5000, perf_data.perf_print_callback)
 
 
     print("Starting pipeline \n")

apps/deepstream-imagedata-multistream/README

@@ -16,8 +16,8 @@
 ################################################################################
 
 Prerequisites:
-- DeepStreamSDK 6.0.1
-- Python 3.6
+- DeepStreamSDK 6.1
+- Python 3.8
 - Gst-python
 - NumPy package
 - OpenCV package