Merge pull request #1290 from prominenceai/v0.31.a.alpha

v0.31.a.alpha

Makefile

@@ -39,7 +39,7 @@ else
 endif
 
 CXX_VERSION:=c++17
-DSL_VERSION:='L"v0.31.alpha"'
+DSL_VERSION:='L"v0.31.a.alpha"'
 
 GLIB_VERSION:=2.0
 GSTREAMER_VERSION:=1.0

README.md

@@ -50,30 +50,6 @@ $ sudo cp /usr/local/lib/librdkafka* /opt/nvidia/deepstream/deepstream-7.0/lib
 ```
 See [Error in DeepStream 7.0 installation instructions - symlink fails to create](https://forums.developer.nvidia.com/t/error-in-deepstream-7-0-installation-instructions-symlink-fails-to-create/296026) for more information.
 
-## -- Release Highlights --
-
-The latest release is [v0.31.alpha](/Release%20Notes/v0.31.alpha.md) 
-
-### New Component Queue management and monitoring services
-All DSL Pipeline Components use a GStream [queue plugin](https://gstreamer.freedesktop.org/documentation/coreelements/queue.html?gi-language=c) to create a new thread boundry for processing. [New services](/docs/api-component.md#component-queue-management) have been added to monitor and control the queue's leaky, current-level, max-size, and min-threshold properties. Callbacks have been added for notification of queue underrun and overrun conditions. 
-
-### New Custom Sources and Sinks
-New services have been added to create [Custom Video Sources](/docs/api-source.md#custom-video-sources) and [Custom Video Sinks](/docs/api-sink.md#custom-video-sinks) Components using publicly released or proprietary GStreamer plugins.
-
-### New DSL GST Caps Objects
-GStreamer caps objects can be created to filter caps for [Custom GST Elements]() used by the [Custom Video Sources](/docs/api-source.md#custom-video-sources), [Custom Components](/docs/api-component.md#custom-components), and [Custom Video Sinks](/docs/api-sink.md#custom-video-sinks)
-
-### All Encode Sinks now support software encoding
-**IMPORTANT** this is a breaking change.
-
-The [File Sink](/docs/api-sink.md#dsl_sink_file_new), [Record Sink](/docs/api-sink.md#dsl_sink_record_new), [RTSP Server Sink](/docs/api-sink.md#dsl_sink_rtsp_server_new), and [WebRTC Sink](/docs/api-sink.md#dsl_sink_webrtc_new) now support five types of encoders:
-* two hardware; H.264, H.265.
-* and now three software; H.264, H.265, and MP4.
-
-The [RTMP Sink](/docs/api-sink.md#dsl_sink_rtmp_new) now supports software and hardware H.264 encoding.
-
-### Extensive memory leak testing has been done.
-All (minor) issues found have been resolved.
 
 ## Contributing
 
@@ -130,8 +106,17 @@ Come join us on [Discord](https://discord.gg/MJvY9jjpAK), an informal place to c
   * [Message Broker](/docs/api-msg-broker.md)
   * [Info API](/docs/api-info.md)
 * [Examples](/docs/examples.md)
-  * [C/C++](/docs/examples-cpp.md)
-  * [Python](/docs/examples-python.md)
+  * [Basic Inference Pipelies](/docs/examples-basic-pipelines.md)
+  * [Multiple Sources, Tilers, and Demuxers](/docs/examples-sources-tiler-demuxer.md)
+  * [Advanced Inference Pipelies](/docs/examples-advanced-pipelines.md)
+  * [Smart Recording](/docs/examples-smart-recording.md)
+  * [Object Detection Event (ODE) Services](/docs/examples-ode-services.md)
+  * [Encoding Frames to JPEG](/docs/examples-encode-and-save-frame.md)
+  * [Dewarping and Segmentation](/docs/examples-dewarping-and-segmentation.md)
+  * [Dynamic Pipelines](/docs/examples-dynamic-pipelines.md)
+  * [Custom Components](/docs/examples-custom-components.md)
+  * [Working with OpenCV](/docs/examples-opencv.md)
+  * [Diagnostics and Utilites](/docs/examples-diagnaostics-and-utilities.md)
   * [Tkinter Reference App](/docs/examples-tkinter.md)
   * [HTML WebRTC Client](/docs/examples-webrtc-html.md)
 * [Using VS Code](/docs/vscode.md)

Release Notes/dsl-releases.md

@@ -2,6 +2,7 @@
 
 | Release                                                     | Date        |
 | ----------------------------------------------------------- | ----------- |
+| [v0.31.a.alpha (patch)](/Release%20Notes/v0.31.a.alpha.md)  | 09/16/2024  |
 | [v0.31.alpha](/Release%20Notes/v0.3`.alpha.md)              | 09/04/2024  |
 | [v0.30.b.alpha (patch)](/Release%20Notes/v0.30.b.alpha.md)  | 08/28/2024  |
 | [v0.30.a.alpha (patch)](/Release%20Notes/v0.30.a.alpha.md)  | 07/14/2024  |

Release Notes/v0.31.a.alpha.md

@@ -0,0 +1,8 @@
+# v0.31.a.alpha (patch) Release Notes
+**Important!**
+* `v0.31.a.alpha` is a **patch** release (patch `a` for the `v0.31.alpha` release).
+* Only the documentation and examples have been updated (and improved).
+* libdsl.so has not changed in this release.
+
+## Issues closed in this release
+*  Cleanup, catagorize, and document all DSL Python and C/C++ Examples [#1288](https://github.com/prominenceai/deepstream-services-library/issues/1288)

docs/examples-advanced-pipelines.md

@@ -0,0 +1,118 @@
+# Advanced Inference Pipelies
+This page documents the following "Advance Inference Pipelines" consiting of
+* [Parallel Inference on Selective Streams](#parallel-inference-on-selective-streams)
+* [Multiple Pipelines Running in Their Own Thread](#multiple-pipelines-running-in-their-own-thread)
+* [Multiple Pipelines with Interpipe Source listening to Pipeline with InterpipeSink](#multiple-pipelines-with-interpipe-source-listening-to-pipeline-with-interpipe-sink)
+* [Single Pipeline with Interpipe Source switching between Multiple Pipelines/Sinks](#single-pipeline-with-interpipe-source-switching-between-multiple-pipelinessinks)
+
+<br>
+
+---
+
+### Parallel Inference on Selective Streams
+
+* [`parallel_inference_on_selective_streams.py`](/examples/python/parallel_inference_on_selective_streams.py)
+* [`parallel_inference_on_selective_streams.cpp`](/examples/cpp/parallel_inference_on_selective_streams.cpp)
+
+```python
+#
+# This example shows how to use a Remuxer Component to create parallel branches,
+# each with their own Inference Components (Preprocessors, Inference Engines, 
+# Trackers, for example). 
+# IMPORTANT! All branches are (currently) using the same model engine and config.
+# files, which is not a valid use case. The actual inference components and 
+# models to use for any specific use cases is beyond the scope of this example. 
+#
+# Each Branch added to the Remuxer can specify which streams to process or
+# to process all. Use the Remuxer "branch-add-to" service to add to specific streams.
+#
+#    stream_ids = [0,1]
+#    dsl_remuxer_branch_add_to('my-remuxer', 'my-branch-0', 
+#        stream_ids, len[stream_ids])
+#
+# You can use the "branch-add" service if adding to all streams
+#
+#    dsl_remuxer_branch_add('my-remuxer', 'my-branch-0')
+# 
+# In this example, 4 RTSP Sources are added to the Pipeline:
+#   - branch-1 will process streams [0,1]
+#   - branch-2 will process streams [1,2]
+#   - branch-3 will process streams [0,2,3]
+#
+# Three ODE Instance Triggers are created to trigger on new object instances
+# events (i.e. new tracker ids). Each is filtering on a unique class-i
+# (vehicle, person, and bicycle). 
+#
+# The ODE Triggers are added to an ODE Handler which is added to the src-pad
+# (output) of the Remuxer.
+#
+# A single ODE Print Action is created and added to each Trigger (shared action).
+# Using multiple Print Actions running in parallel -- each writing to the same 
+# stdout buffer -- will result in the printed data appearing interlaced. A single 
+# Action with an internal mutex will protect from stdout buffer reentrancy. 
+# 
+
+```
+
+<br>
+
+---
+
+### Multiple Pipelines Running in Their Own Thread
+
+* [`multiple_pipelines.py`](/examples/python/multiple_pipelines.py)
+* [`multiple_pipelines.cpp`](/examples/cpp/multiple_pipelines.cpp)
+
+```python
+#
+# This example demonstrates how to run multple Pipelines, each in their own 
+# thread, and each with their own main-context and main-loop.
+#
+# After creating and starting each Pipelines, the script joins each of the 
+# threads waiting for them to complete - either by EOS message, 'Q' key, or 
+# Delete Window.
+#
+```
+
+<br>
+
+---
+
+### Multiple Pipelines with Interpipe Source listening to Pipeline with Interpipe Sink
+
+* [`interpipe_multiple_pipelines_listening_to_single_sink.py`](/examples/python/interpipe_multiple_pipelines_listening_to_single_sink.py)
+* [`interpipe_multiple_pipelines_listening_to_single_sink.cpp`](/examples/cpp/interpipe_multiple_pipelines_listening_to_single_sink.cpp)
+
+```python
+#
+# This script demonstrates how to run multple Pipelines, each with an Interpipe
+# Source, both listening to the same Interpipe Sink.
+#
+# A single Player is created with a File Source and Interpipe Sink. Two Inference
+# Pipelines are created to listen to the single Player - shared input stream. 
+# The two Pipelines can be created with different configs, models, and/or Trackers
+# for side-by-side comparison. Both Pipelines run in their own main-loop with their 
+# own main-context, and have their own Window Sink for viewing and external control.
+#
+```
+<br>
+
+---
+
+### Single Pipeline with Interpipe Source switching between Multiple Pipelines/Sinks
+
+* [`interpipe_single_pipeline_dynamic_switching_between_multiple_sinks.py`](/examples/python/interpipe_single_pipeline_dynamic_switching_between_multiple_sinks.py)
+* [`interpipe_single_pipeline_dynamic_switching_between_multiple_sinks.cpp`](/examples/cpp/interpipe_single_pipeline_dynamic_switching_between_multiple_sinks.cpp)
+
+```python
+# ------------------------------------------------------------------------------------
+# This example demonstrates interpipe dynamic switching. Four DSL Players
+# are created, each with a File Source and Interpipe Sink. A single
+# inference Pipeline with an Interpipe Source is created as the single listener
+# 
+# The Interpipe Source's "listen_to" setting is updated based on keyboard input.
+# The xwindow_key_event_handler (see below) is added to the Pipeline's Window Sink.
+# The handler, on key release, sets the "listen_to" setting to the Interpipe Sink
+# name that corresponds to the key value - 1 through 4.
+```
+