# Deepstream Smart Record Python **URL:** \ **Category:** DeepStream SDK\ **Tags:** deepstream\ **Created:** [July 10, 2025, 9:11am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587 "2025-07-10T09:11:39Z")\ **Posts on this page:** 11\ **Page:** 1
**Author:** ![jj94](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@jj94](https://forums.developer.nvidia.com/u/jj94)\ **Post date:** [July 10, 2025, 9:11am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/1 "2025-07-10T09:11:39Z")
Please provide complete information as applicable to your setup. **• Hardware Platform (Jetson / GPU)** dGPU 3070 ti laptop **• DeepStream Version** 7.1 **• JetPack Version (valid for Jetson only)** **• TensorRT Version** 10.12 **• NVIDIA GPU Driver Version (valid for GPU only)** **• Issue Type( questions, new requirements, bugs)** **• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)** **• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)** I followed the steps for compiling python bindings with smart record enabled from [DeepStream SDK FAQ - #41 by fanzh](https://forums.developer.nvidia.com/t/deepstream-sdk-faq/80236/41)[nvurisrcbin plugin to perform smart recording in] I then tried to replicate the test1 pipeline , but it starts the recording but never stops it\> I am attaching my code and output. Code: ```auto nvurisrcbin = None def stop_record_callback(): global nvurisrcbin nvurisrcbin.emit('stop-sr', 0) # keep same as native for get size of native struct class SRUserContext(ctypes.Structure): _fields_ = [ ("sessionid", ctypes.c_int), ("name", ctypes.c_char * 32) ] # GstElement * src, NvDsSRRecordingInfo* recordingInfo, void* data, void* user_data def record_done(nvurisrcbin, recordingInfo, user_ctx, user_data): print(' ******sr done*****') print(f"record_done nvurisrcbin {nvurisrcbin} recordingInfo {recordingInfo}" f" user_ctx {user_ctx} user_data {user_data}") info = pyds.NvDsSRRecordingInfo.cast(hash(recordingInfo)) sr = pyds.SRUserContext.cast(hash(user_ctx)) print(f"filename {info.filename} -- dir {info.dirpath} {info.width} x {info.height}") print(f"session id {sr.sessionid} -- name {sr.name}") ``` ```auto global nvurisrcbin nvurisrcbin = Gst.ElementFactory.make("nvurisrcbin", "uri-decode-bin") if not nvurisrcbin: sys.stderr.write(" Unable to create nvurisrcbin \n") nvurisrcbin.set_property("smart-record", 2) # enable smart record. nvurisrcbin.set_property("smart-rec-dir-path", "./") # set record path. # Create nvstreammux instance to form batches from one or more sources. streammux = Gst.ElementFactory.make("nvstreammux", "Stream-muxer") if not streammux: sys.stderr.write(" Unable to create NvStreamMux \n") nvurisrcbin.set_property("uri",uri) # Connect to the "pad-added" signal of the decodebin which generates a # callback once a new pad for raw data has beed created by the decodebin nvurisrcbin.connect("pad-added", cb_newpad, streammux) nvurisrcbin.connect("sr-done", record_done, pipeline) ``` ```auto pipeline.set_state(Gst.State.PLAYING) sessionid = pyds.get_native_ptr(pyds.alloc_buffer(4)) print(f"sessionid {sessionid}") sr_user_context_size = ctypes.sizeof(SRUserContext) sr_user_context_buf = pyds.get_native_ptr(pyds.alloc_buffer(sr_user_context_size)) sr_user_context = pyds.SRUserContext.cast(sr_user_context_buf) sr_user_context.sessionid = 42 sr_user_context.name = "sr-demo" print(f"sr_user_context_buf {sr_user_context_buf}") # signal_info = GObject.signal_query('start-sr', nvurisrcbin) # print(f"signal_info {signal_info.param_types}") # stop smart record manually, optional # GLib.timeout_add(5000, stop_record_callback) def start_sr_function(): nvurisrcbin.emit('start-sr', sessionid, 2, 5, sr_user_context_buf) print(' ******start sr*****') return True timer_id = GLib.timeout_add_seconds(10, start_sr_function) try: loop.run() except: pass # cleanup pipeline.set_state(Gst.State.NULL) # release native memory pyds.free_gbuffer(sessionid) pyds.free_gbuffer(sr_user_context_buf) ``` Output: ```auto ❯ uv run test Creating Pipeline Is it Integrated GPU? : 0 Creating EGLSink Playing file{clip} Adding elements to Pipeline Linking elements in the Pipeline Starting pipeline 0:00:00.274042625 479665 0x210a9210 INFO nvinfer gstnvinfer.cpp:684:gst_nvinfer_logger: NvDsInferContext[UID 1]: Info from NvDsInferContextImpl::deserializeEngineAndBackend() [UID = 1]: deserialized trt engine from :/home/jagpreet/weapon_detection/model/resnet18_trafficcamnet_pruned.onnx_b1_gpu0_int8.engine Implicit layer support has been deprecated INFO: ../nvdsinfer/nvdsinfer_model_builder.cpp:327 [Implicit Engine Info]: layers num: 0 0:00:00.274084395 479665 0x210a9210 INFO nvinfer gstnvinfer.cpp:684:gst_nvinfer_logger: NvDsInferContext[UID 1]: Info from NvDsInferContextImpl::generateBackendContext() [UID = 1]: Use deserialized engine model: /home/jagpreet/weapon_detection/model/resnet18_trafficcamnet_pruned.onnx_b1_gpu0_int8.engine 0:00:00.281561824 479665 0x210a9210 INFO nvinfer gstnvinfer_impl.cpp:343:notifyLoadModelStatus: [UID 1]: Load new model:test1_pgie.txt sucessfully sessionid sr_user_context_buf Failed to query video capabilities: Invalid argument In cb_newpad gstname= video/x-raw features= ******start sr***** ******start sr***** ******start sr***** ******start sr***** ******start sr***** nvstreammux: Successfully handled EOS for source_id=0 End-of-stream ``` [out.txt](https://forums.developer.nvidia.com/uploads/short-url/pcq8aYDCYWNJfXEEg8fygQEQ7FY.txt) (1.7 KB) [test.txt](https://forums.developer.nvidia.com/uploads/short-url/rAaNuX9ZtRqMY7iUi1mQ2l6TwUq.txt) (12.4 KB) ---
**Author:** ![fanzh](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/fanzh/32/115955_2.png) [@fanzh](https://forums.developer.nvidia.com/u/fanzh)\ **Post date:** [July 10, 2025, 2:26pm UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/3 "2025-07-10T14:26:56Z")
In your code, “stop-sr” signal is not sent. you can send “stop-sr” to stop the recording, Or the recording will stop automatically after the recording duration is bigger than the SMART\_REC\_DEFAULT\_DURATION. Please refer to the C version sample deepstream-testsr. ---
**Author:** ![jj94](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@jj94](https://forums.developer.nvidia.com/u/jj94)\ **Post date:** [July 10, 2025, 3:54pm UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/4 "2025-07-10T15:54:14Z")
I tried adding it like this: ```auto def start_sr_function(): nvurisrcbin.emit('start-sr', sessionid, 1, 5, sr_user_context_buf) print(' ******start sr*****') GLib.timeout_add_seconds(15, stop_record_callback) return False ``` Also in output I get ```auto sessionid sr_user_context_buf Failed to query video capabilities: Invalid argument In cb_newpad gstname= video/x-raw features= ******start sr***** ``` are session\_id and sr\_user\_context\_buf supposed to be NULL?? ---
**Author:** ![fanzh](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/fanzh/32/115955_2.png) [@fanzh](https://forums.developer.nvidia.com/u/fanzh)\ **Post date:** [July 11, 2025, 2:05am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/5 "2025-07-11T02:05:40Z")
1. No, get\_native\_ptr is opensource. it only prints the “native\_ptr”. 2. Please add log in stop\_record\_callback. do you mean “_sr done_” is not printed after stop\_record\_callback is called? if so, was the recording generated? is the same issue with ths [topic](https://forums.developer.nvidia.com/t/nvidia-ds-smart-recording-callback-is-not-working/334364)? ---
**Author:** ![jj94](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@jj94](https://forums.developer.nvidia.com/u/jj94)\ **Post date:** [July 11, 2025, 6:11am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/6 "2025-07-11T06:11:31Z")
I added log to the stop callback: ```auto def stop_record_callback(): global nvurisrcbin nvurisrcbin.emit('stop-sr', 0) print("Sr stop cb called") ``` Output: ```auto gstname= video/x-raw features= ******start sr***** ******start sr***** Sr stop cb called ******start sr***** Sr stop cb called ******start sr***** Sr stop cb called nvstreammux: Successfully handled EOS for source_id=0 End-of-stream ``` No recording is generated. and record\_done is not called either. I also ran `gst-launch-1.0 nvurisrcbin uri=file:///xxx.mp4 ! fakesink` from the linked topic, output: ```auto Pipeline is PREROLLED ... Setting pipeline to PLAYING ... Redistribute latency... New clock: GstSystemClock Redistribute latency... Got EOS from element "pipeline0". Execution ended after 0:00:02.038409674 Setting pipeline to NULL ... Freeing pipeline ... ``` I am also attaching my bindfuctions.cpp and utils.cpp [bindfunc.txt](https://forums.developer.nvidia.com/uploads/short-url/bfZ6JHwkS1Ghq6BHEMRq0nVFi2C.txt) (35.9 KB) [utils.txt](https://forums.developer.nvidia.com/uploads/short-url/awzBATGwy50UtCI1ZdihVO31cxh.txt) (9.1 KB) I made the changes you made in [**Use nvurisrcbin plugin to do smart record in Python**](https://forums.developer.nvidia.com/t/deepstream-sdk-faq/80236/41) But I also saw in another post “cpp patch (`gpointer_bind.dff` )” which adds alloc\_buffer1 on bindfunctions.cpp and functiondoc.h. Do I need to apply that patch as well? and use allocbuffer1 instead of allocbuffer? ---
**Author:** ![fanzh](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/fanzh/32/115955_2.png) [@fanzh](https://forums.developer.nvidia.com/u/fanzh)\ **Post date:** [July 11, 2025, 6:37am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/7 "2025-07-11T06:37:34Z")
1. As the [doc](https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_plugin_gst-nvurisrcbin.html) shown, “Sources must be of type source-type-rtsp” , meaning nvurisrcbin smart recording only support RTSP source. 2. Please only refer to the FAQ [**Use nvurisrcbin plugin to do smart record in Python**](https://forums.developer.nvidia.com/t/deepstream-sdk-faq/80236/41). some topics are too old. ---
**Author:** ![jj94](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@jj94](https://forums.developer.nvidia.com/u/jj94)\ **Post date:** [July 11, 2025, 8:58am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/8 "2025-07-11T08:58:42Z")
Thank you so much, that solved the smart record problem. However I get a warning when running with the rtsp: ```auto There may be a timestamping problem, or this computer is too slow. Warning: gst-core-error-quark: A lot of buffers are being dropped. (13): ../libs/gst/base/gstbasesink.c(3147): gst_base_sink_is_too_late (): /GstPipeline:pipeline0/GstEglGlesSink:nvvideo-renderer: There may be a timestamping problem, or this computer is too slow. ******start sr***** ** WARN: : Invalid PTS found in stream. Stream recorded by Smart Record might have issues Warning: gst-core-error-quark: A lot of buffers are being dropped. (13): ../libs/gst/base/gstbasesink.c(3147): gst_base_sink_is_too_late (): /GstPipeline:pipeline0/GstEglGlesSink:nvvideo-renderer: There may be a timestamping problem, or this computer is too slow. Warning: gst-core-error-quark: A lot of buffers are being dropped. (13): ../libs/gst/base/gstbasesink.c(3147): gst_base_sink_is_too_late (): /GstPipeline:pipeline0/GstEglGlesSink:nvvideo-renderer: ``` But recording works fine ---
**Author:** ![fanzh](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/fanzh/32/115955_2.png) [@fanzh](https://forums.developer.nvidia.com/u/fanzh)\ **Post date:** [July 11, 2025, 9:03am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/9 "2025-07-11T09:03:47Z")
if you are using eglsink, please set the sync property to false, meaning playing as fast as possible. Or, you can use fakesink instead. ---
**Author:** ![jj94](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@jj94](https://forums.developer.nvidia.com/u/jj94)\ **Post date:** [July 11, 2025, 9:09am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/10 "2025-07-11T09:09:26Z")
that solved it, ```auto gstname= video/x-raw features= Failed to query video capabilities: Invalid argument ******start sr***** ** WARN: : Invalid PTS found in stream. Stream recorded by Smart Record might have issues Failed to query video capabilities: Invalid argument ******sr done***** record_done nvurisrcbin < __gi__.GstDsNvUriSrcBin object at 0x75d2a1933780 (GstDsNvUriSrcBin at 0x10584600)> recordingInfo user_ctx user_data filename 129540978380160 -- dir 297277760 2560 x 1920 session id 42 -- name sr-demo ******start sr***** ** ERROR: : No video stream found ******start sr***** ** ERROR: : No video stream found ******start sr***** ``` One Last Question Can I trigger the sr start using a probe after nvinfer?? and What if Want to save the video with overlayed bboxes in case of using a detection model or label when using a classification model? ---
**Author:** ![fanzh](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/fanzh/32/115955_2.png) [@fanzh](https://forums.developer.nvidia.com/u/fanzh)\ **Post date:** [July 11, 2025, 9:19am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/11 "2025-07-11T09:19:11Z")
yes, you can send the ‘start’ signal in probe function. if you want to record the video with bboxes, you need to do encoding after nvosd plugin. please refer to the ready-made sample deepstream-testsr in DeepStream SDK. nvurisrcbin only supports recording the RTSP stream. ---
**Author:** ![system](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/system/32/68080_2.png) [@system](https://forums.developer.nvidia.com/u/system)\ **Post date:** [July 25, 2025, 9:19am UTC](https://forums.developer.nvidia.com/t/deepstream-smart-record-python/338587/13 "2025-07-25T09:19:16Z")
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.