Merge pull request #1302 from prominenceai/v0.31.b.alpha

v0.31.b.alpha

Makefile

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

Release Notes/dsl-releases.md

@@ -2,6 +2,7 @@
 
 | Release                                                     | Date        |
 | ----------------------------------------------------------- | ----------- |
+| [v0.31.b.alpha (patch)](/Release%20Notes/v0.31.b.alpha.md)  | 11/02/2024  |
 | [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  |

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

@@ -0,0 +1,20 @@
+# v0.31.b.alpha (patch) Release Notes
+**Important!**
+* `v0.31.b.alpha` is a **patch** release (patch `b` for the `v0.31.alpha` release).
+
+## Purpose
+The purpose for this patch release:
+1. Remove the invalid guard preventing dynamic ["on-the-fly"](https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_on_the_fly_model.html) model engine updates. Support dynamicl model engine updates with a new "update-complete" listener callback function.  
+2. Update the "Add IoT Message Meta Action" to add the Trigger name to identify the source of the event. 
+
+## Issues-bugs closed in this release
+* The GIE and TIS "config-file-path" and "model-engine-file" properties should be writable in any state [#1295](https://github.com/prominenceai/deepstream-services-library/issues/1295)
+*  Add IoT Message Meta Action must add the Trigger Name in the NvDsEventMsgMeta to identify the event source. [#1298](https://github.com/prominenceai/deepstream-services-library/issues/1298)
+
+## Issues-enhancements closed in this release
+* Implement dsl_infer_gie_model_update_listener_add/remove services for async model update notifications [#1297](https://github.com/prominenceai/deepstream-services-library/issues/1297)
+*  Implement new dynamic "on-the-fly" model-engine update examples using new update-listener callback services. [#1299](https://github.com/prominenceai/deepstream-services-library/issues/1297)
+
+## New Examples in this release
+* [dynamically_update_inference_model.py](/examples/python/dynamically_update_inference_model.py)
+* [dynamically_update_inference_model.cpp](/examples/cpp/dynamically_update_inference_model.cpp)

docs/api-reference-list.md

@@ -30,6 +30,7 @@
 * [`dsl_sink_window_key_event_handler_cb`](/docs/api-sink.md#dsl_sink_window_key_event_handler_cb)
 * [`dsl_sink_window_button_event_handler_cb`](/docs/api-sink.md#dsl_sink_window_button_event_handler_cb)
 * [`dsl_sink_window_delete_event_handler_cb`](/docs/api-sink.md#dsl_sink_window_delete_event_handler_cb)
+* [`dsl_infer_gie_model_update_listener_cb`](/docs/api-infer.md#dsl_infer_gie_model_update_listener_cb)
 
 ## DSL Services API:
 * [`dsl_main_loop_run`](/docs/overview.md#main-loop-context)
@@ -308,13 +309,15 @@
 * [`dsl_infer_gie_secondary_new`](/docs/api-infer.md#dsl_infer_gie_secondary_new)
 * [`dsl_infer_tis_primary_new`](/docs/api-infer.md#dsl_infer_tis_primary_new)
 * [`dsl_infer_tis_secondary_new`](/docs/api-infer.md#dsl_infer_tis_secondary_new)
-* [`dsl_infer_batch_size_get`](/docs/api-infer.md#dsl_infer_batch_size_get)
-* [`dsl_infer_batch_size_set`](/docs/api-infer.md#dsl_infer_batch_size_set)
-* [`dsl_infer_unique_id_get`](/docs/api-infer.md#dsl_infer_unique_id_get)
 * [`dsl_infer_gie_model_engine_file_get`](/docs/api-infer.md#dsl_infer_gie_model_engine_file_get)
 * [`dsl_infer_gie_model_engine_file_set`](/docs/api-infer.md#dsl_infer_gie_model_engine_file_set)
+* [`dsl_infer_gie_model_update_listener_add`](/docs/api-infer.md#dsl_infer_gie_model_update_listener_add)
+* [`dsl_infer_gie_model_update_listener_remove`](/docs/api-infer.md#dsl_infer_gie_model_update_listener_remove)
 * [`dsl_infer_gie_tensor_meta_settings_get`](/docs/api-infer.md#dsl_infer_gie_tensor_meta_settings_get)
 * [`dsl_infer_gie_tensor_meta_settings_set`](/docs/api-infer.md#dsl_infer_gie_tensor_meta_settings_set)
+* [`dsl_infer_batch_size_get`](/docs/api-infer.md#dsl_infer_batch_size_get)
+* [`dsl_infer_batch_size_set`](/docs/api-infer.md#dsl_infer_batch_size_set)
+* [`dsl_infer_unique_id_get`](/docs/api-infer.md#dsl_infer_unique_id_get)
 * [`dsl_infer_config_file_get`](/docs/api-infer.md#dsl_infer_config_file_get)
 * [`dsl_infer_config_file_set`](/docs/api-infer.md#dsl_infer_config_file_set)
 * [`dsl_infer_interval_get`](/docs/api-infer.md#dsl_infer_interval_get)

docs/examples-basic-pipelines.md

@@ -4,6 +4,7 @@ This page documents the following "Basic Inference Pipelines" consiting of
 * [File Source, Primary GIE, IOU Tracker, OSD, EGL Window Sink, and File Sink](#file-source-primary-gie-iou-tracker-osd-egl-window-sink-and-file-sink)
 * [File Source, Primary GIE, IOU Tracker, OSD, EGL Window Sink, and RTSP Sink](#file-source-primary-gie-iou-tracker-osd-egl-window-sink-and-rtsp-sink)
 * [File Source, Primary GIE, IOU Tracker, OSD, EGL Window Sink, and V4L2 Sink](#file-source-primary-gie-iou-tracker-osd-egl-window-sink-and-v4l2-sink)
+* [File Source, Primary GIE, DCF Tracker, 2 Secondary GIEs, OSD, EGL Windon Sink](#file-source-primary-gie-dcf-tracker-2-secondary-gies-osd-egl-windon-sink)
 * [RTSP Source, Primary GIE, IOU Tracker, OSD, EGL Window Sink](#rtsp-source-primary-gie-iou-tracker-osd-egl-window-sink)
 * [HTTP Source, Primary GIE, IOU Tracker, OSD, EGL Window Sink](#http-source-primary-gie-iou-tracker-osd-egl-window-sink)
 * [File Source, Preprocessor, Primary GIE, IOU Tracker, OSD, EGL Window Sink](#file-source-preprocessor-primary-gie-iou-tracker-osd-egl-window-sink)
@@ -139,6 +140,33 @@ This page documents the following "Basic Inference Pipelines" consiting of
 
 ---
 
+### File Source, Primary GIE, DCF Tracker, 2 Secondary GIEs, OSD, EGL Windon Sink
+* [1file_pgie_dcf_tracker_2sgie_window.py](/examples/python/1file_pgie_dcf_tracker_2sgie_window.py)
+* [1file_pgie_dcf_tracker_2sgie_window.cpp](/examples/cpp/1file_pgie_dcf_tracker_2sgie_window.cpp)
+
+```python
+#
+# The simple example demonstrates how to create a set of Pipeline components, 
+# specifically:
+#   - File Source
+#   - Primary GST Inference Engine (PGIE)
+#   - DCF Tracker
+#   - 2 Secondary GST Inference Engines (SGIEs)
+#   - On-Screen Display (OSD)
+#   - Window Sink
+# ...and how to add them to a new Pipeline and play
+# 
+# The example registers handler callback functions with the Pipeline for:
+#   - key-release events
+#   - delete-window events
+#   - end-of-stream EOS events
+#   - Pipeline change-of-state events
+#  
+```
+<br>
+
+---
+
 ### RTSP Source, Primary GIE, IOU Tracker, OSD, EGL Window Sink
 
 * [`1rtsp_pgie_dcf_tracker_osd_window.py`](/examples/python/1rtsp_pgie_dcf_tracker_osd_window.py)

docs/examples-dynamic-pipelines.md

@@ -1,12 +1,61 @@
 # Dynamic Pipelines
 This page documents the following examples:
+* [Dynamically Update Inference Model Engine File](#dynamically-update-inference-model-engine-file)
 * [Dynamically Add/Remove Sources to/from a Pipeline with a Tiler and Window Sink](#dynamically-addremove-sources-tofrom-a-pipeline-with-a-tiler-and-window-sink)
 * [Dynamically Move a Branch from One Demuxer Stream to Another](#dynamically-move-a-branch-from-one-demuxer-stream-to-another)
 
 <br>
 
 ---
 
+### Dynamically Update Inference Model Engine File
+* [dynamically_update_inference_model.py](/examples/python/dynamically_update_inference_model.py)
+* [dynamically_update_inference_model.cpp](/examples/cpp/dynamically_update_inference_model.cpp)
+
+```python
+#
+# The simple example demonstrates how to create a set of Pipeline components, 
+# specifically:
+#   - File Source
+#   - Primary GST Inference Engine (PGIE)
+#   - DCF Tracker
+#   - Secondary GST Inference Engines (SGIEs)
+#   - On-Screen Display (OSD)
+#   - Window Sink
+# ...and how to dynamically update an Inference Engine's config and model files.
+#  
+# The key-release handler function will dynamically update the Secondary
+# Inference Engine's config-file on the key value as follows.
+#
+#    "1" = '../../test/config/config_infer_secondary_vehicletypes.yml'
+#    "2" = '../../test/config/config_infer_secondary_vehiclemake.yml'
+#
+# The new model engine is loaded by the SGIE asynchronously. a client listener 
+# (callback) function is added to the SGIE to be notified when the loading is
+# complete. See the "model_update_listener" function defined below.
+#   
+# IMPORTANT! it is best to allow the config file to specify the model engine
+# file when updating both the config and model. Set the model_engine_file 
+# parameter to None when creating the Inference component.
+#  
+#        retval = dsl_infer_gie_secondary_new(L"secondary-gie", 
+#            secondary_infer_config_file_1.c_str(), NULL, L"primary-gie", 0);
+#
+# The Config files used are located under /deepstream-services-library/test/config
+# The files reference models created with the file 
+#     /deepstream-services-library/make_trafficcamnet_engine_files.py
+# 
+# The example registers handler callback functions with the Pipeline for:
+#   - key-release events
+#   - delete-window events
+#   - end-of-stream EOS events
+#   - Pipeline change-of-state events#   
+#
+```
+<br>
+
+---
+
 ### Dynamically Add/Remove Sources to/from a Pipeline with a Tiler and Window Sink
 
 * [`dynamically_add_remove_sources_with_tiler_window_sink.py`](/examples/python/dynamically_add_remove_sources_with_tiler_window_sink.py)

dsl.py

@@ -562,6 +562,10 @@ class dsl_threshold_value(Structure):
 DSL_COMPONENT_QUEUE_UNDERRUN_LISTENER = \
     CFUNCTYPE(None, c_wchar_p, c_void_p)
 
+#dsl_infer_gie_model_update_listener_cb
+DSL_INFER_GIE_MODEL_UPDATE_LISTENER = \
+    CFUNCTYPE(None, c_wchar_p, c_wchar_p, c_void_p)
+
 ##
 ## TODO: CTYPES callback management needs to be completed before any of
 ## the callback remove wrapper functions will work correctly.
@@ -5124,6 +5128,35 @@ def dsl_infer_raw_output_enabled_set(name, enabled, path):
     result = _dsl.dsl_infer_raw_output_enabled_set(name, enabled, path)
     return int(result)
 
+##
+## dsl_infer_gie_model_update_listener_add()
+##
+_dsl.dsl_infer_gie_model_update_listener_add.argtypes = [c_wchar_p, 
+    DSL_INFER_GIE_MODEL_UPDATE_LISTENER, c_void_p]
+_dsl.dsl_infer_gie_model_update_listener_add.restype = c_uint
+def dsl_infer_gie_model_update_listener_add(name, listener, client_data):
+    global _dsl
+    c_listener = DSL_INFER_GIE_MODEL_UPDATE_LISTENER(listener)
+    callbacks.append(c_listener)
+    c_client_data=cast(pointer(py_object(client_data)), c_void_p)
+    clientdata.append(c_client_data)
+    result = _dsl.dsl_infer_gie_model_update_listener_add(name, 
+        c_listener, c_client_data)
+    return int(result)
+
+##
+## dsl_infer_gie_model_update_listener_remove()
+##
+_dsl.dsl_infer_gie_model_update_listener_remove.argtypes = [c_wchar_p, 
+    DSL_INFER_GIE_MODEL_UPDATE_LISTENER]
+_dsl.dsl_infer_gie_model_update_listener_remove.restype = c_uint
+def dsl_infer_gie_model_update_listener_remove(name, listener):
+    global _dsl
+    c_listener = DSL_INFER_GIE_MODEL_UPDATE_LISTENER(listener)
+    result = _dsl.dsl_infer_gie_model_update_listener_remove(name, c_listener)
+    return int(result)
+
+
 ##
 ## dsl_tracker_new()
 ##