Date   

Re: LF AI&Data NNstreamer Project License Scan and Findings July 2022

MyungJoo Ham
 

Finding #7 (Pytorch devel package has GPL w/ GCC runtime exception headers).won't be fixed by removing the corresponding header files. The header files with GPL issue are essential headers of PyTorch, which is required to build C/C++ code that uses PyTorch.

Thus, we will make sure that the module that uses these (nnstreamer-pytorch plugin) is compiled by GCC. It will generate build errors if nnstreamer-pytorch is not being built by GCC.


Re: LF AI&Data NNstreamer Project License Scan and Findings July 2022

MyungJoo Ham
 

Most of the issues are coming from "nnshark", a fork of "gstshark", which is tuned for pipelines with nnstreamer. It will be handled by Wook although it might be slow.

The issues related with nnstreamer-android-resource (Finding #6, #7) will be handled by Yongjoo ( https://github.com/nnstreamer/nnstreamer/issues/3873 )

Anyway, for Finding #4, other than nnstreamer.git and a few other repositories, which directly inherits LGPL 2.1 code, other repositories that do not have obligations to LGPL are intentionally using Apache 2.0.

Thank you for the report.

Cheers, MyungJoo


LF AI&Data NNstreamer Project License Scan and Findings July 2022

Jeff Shapiro <jshapiro@...>
 

Hi Team,

Here are the results from the July 2022 license scan of the NNstreamer project. The scan was performed using the Linux Foundation Fossology server. Licenses and copyrights were examined.

The key findings (if any) and license summary can be found in the HTML report, the list of files in the spreadsheet, and also find the SPDX file listed below:

REPORTS:

lfai/nnstreamer, code pulled 2022-07-02
- report: https://lfscanning.org/reports/lfai/nnstreamer-2022-07-02-8eb54f0b-f36e-43ff-826f-80e9f251e17e.html
- xlsx: https://lfscanning.org/reports/lfai/nnstreamer-2022-07-02-8eb54f0b-f36e-43ff-826f-80e9f251e17e.xlsx
- spdx: https://github.com/lfscanning/spdx-lfai/tree/master/nnstreamer/2022-07/nnstreamer-2022-07-02.spdx

NOTE: Some of the key findings are old from prior scans, and some are new from this current scan. There are a lot of findings, please take a careful look and let me know if you want to discuss.

Please feel free to contact me with any questions about the scan results. Be sure to reply to me directly as I may not get an email sent directly to the distribution list.

Thanks, Jeff

Jeff Shapiro
408-910-7792
jshapiro@...


Re: LF AI NNstreamer Project Scan and Findings

MyungJoo Ham
 

Finding #1: for the nnshark/gstshark issue, I've contacted gstshark: https://github.com/RidgeRun/gst-shark/issues/105

It is a fork of RidgeRun/gstshark, I'd first ask them to clarify the license.

Finding #4: Yes, there are a few projects intentionally licensed under Apache 2.0; where users are expected to use/update/include it directly, not linking as a shared library. In such cases, LGPL will be a roadblock against commercial embedded devices.

I'll keep looking at other Findings, too.

Thanks a lot!


LF AI NNstreamer Project Scan and Findings

Jeff Shapiro <jshapiro@...>
 

Hi Team,

I'm taking over the licesne scanning responsibilities from Steve Winslow.

Here are the results from the most recent license scan of the NNstreamer project.  The scan was performed on the codebase at:  https://github.com/nnstreamer/nnstreamer  based on a snapshot from Feb 2, using the Linux Foundation Fossology server.  Licenses and copyrights were examined.

The key findings (if any) and license summary can be found in the HTML report, the list of files in the spreadsheet, and also find the SPDX file listed below:

REPORTS:

lfai/nnstreamer, code pulled 2022-02-02
  - report: https://lfscanning.org/reports/lfai/nnstreamer-2022-02-02-944feaca-0a1b-4a63-bf63-72370e333b96.html
  - xlsx:   https://lfscanning.org/reports/lfai/nnstreamer-2022-02-02-944feaca-0a1b-4a63-bf63-72370e333b96.xlsx
  - spdx:   https://github.com/lfscanning/spdx-lfai/tree/master/nnstreamer/2022-01/nnstreamer-2022-02-02.spdx


NOTE:  For any key findings listed, they may be new or they may be carried over from the last scan that Steve did last September.

Please feel free to contact me with any questions about the scan results.  Be sure to reply to me directly as I may not get an email sent directly to the distribution list.

Thanks, Jeff

Jeff Shapiro
408-910-7792
jshapiro@...




Re: nnstreamer build issue on Android studio

MyungJoo Ham
 

  • File access permissions and the privileges of the building processes?
  • If you have nnstreamer aar already, why are you building nnstreamer?

You'll need to find the difference between the two systems.

If you want some meaningful answers, you will need to elaborate

  1. the difference between the two.

  2. effective info on which step did the build fail (which file was failed to be compiled? or linked?)

  3. what does the successful system do with 2.


nnstreamer build issue on Android studio

Sapna Kumari <sapna2602@...>
 

Dear Team,

I am trying to build nnstreamer-ssd example in android studio.
I have followed the steps:
- nnstreamer aar file to api-sample/libs
- tensorflow-lite extracted to common/jni/
- gstreamer and nnstreamer root path specified
- ndk version 21.0.6

In one system, I am able to build the example and generate apk.
But with same settings when trying to build on another system, error comes:
ndk-build failed :
nnstreamer/gst/nnstreamer/hw_accel.o : no such file or directory.

But when i checked in build/intermediates directory, the file is present.

Is there something I am missing?

Please guide.

Thanks,
Sapna


Re: Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

MyungJoo Ham
 

It is general C library and nnstreamer uses the standard building procedure.

You may refer to the nnstreamer documentation; however, please keep in mind that you can refer to the general Linux C library building procedure (meson+ninja).  You may discuss building issues/questions in github-issues, too.

 

--------- Original Message ---------

Sender : 이병헌 <byungs286@...>

Date : 2021-12-02 18:57 (GMT+9)

Title : Re: [NNStreamer Technical Discuss] Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

 

Thanks 

I don't know well about rebuiling procedure of nnstreamer, but I will try with github documentations.
If my problem is resolved, I will record it into this thread

regards.

 

 


Re: Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

이병헌
 

Thanks 

I don't know well about rebuiling procedure of nnstreamer, but I will try with github documentations.
If my problem is resolved, I will record it into this thread

regards.


Re: Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

MyungJoo Ham
 

Ah.. as Jihoon said, if you have a model for PyTorch 1.10, you need to install and run PyTorch 1.10. Rebuild nnstreamer-pytorch with PyTorch 1.10 if your nnstreamer-pytorch installation doesn't work w/ PyTorch 1.10

In other words:

Try 1: keep nnstreamer, nnstreamer-pytorch, install PyTorch 1.10 and let nnstreamer run w/ PyTorch 1.10 (need to make sure the version you use in run-time) : easiest. but not sure if this will work (if PyTorch has changed header files, this will fail)

Try 2: keep nnstreamer, rebuild nnstreamer-pytorch based on PyTorch 1.10. (this should work. requires a bit of C building tricks.)

Try 3: rebuild the whole nnstreamer w/ PyTorch. (this should work. doesn't require C building tricks.)


Re: Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

Jihoon Lee
 

Hello,

Apart from the pipeline description, backward-compatibility should mean that Pytorch 1.10.1 can load model built with pytorch 1.3.1 not vice versa.

(eg) excel 2010 can open file saved with excel 2003 but not the other way around)

If problem persists, you might want to check if libtorch 1.3.1 is capable of loading yolov5s.torchscript.pt or use suitable libtorch version.

 

Bests,

Ji

 

From: nnstreamer-technical-discuss@... <nnstreamer-technical-discuss@...> On Behalf Of 이병헌
Sent: Thursday, December 2, 2021 3:09 PM
To: nnstreamer-technical-discuss@...
Subject: [NNStreamer Technical Discuss] Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

 

Hi dev.

I’m trying to use yolov5s tensorscript.pt file that exported from yolov5 GitHub code with nnstreamer pipeline.

I’m working on develop environment with nnstreamer/tools/docker/ubuntu18.04-run/Dockerfile

 

Below is my test pipeline.

 

const char *string = "rtspsrc location=rtsp://address:port/mount latency=0 \

protocols=4 ! rtph265depay ! avdec_h265 ! \

videoscale ! videoconvert ! video/x-raw,format=RGB,width=640,height=640 ! \

tensor_converter ! tensor_filter framework=pytorch \

model=../../tf_model/yolov5s.torchscript.pt \

input=3:640:640:1 inputname=x inputtype=float32 \

output=1:25200:85 outputname=416 outputtype=float32 \

! tensor_sink name=tensor_sink”;

 

But with this pipeline, I got error as below

 

failed to initialize the object: PyTorch

0:00:00.376511035    29 0x559a1ba3d4c0 WARN                GST_PADS gstpad.c:1149:gst_pad_set_active:<tensorfilter0:sink> Failed to activate pad

** Message: 05:11:55.607: gpu = 0, accl = cpu

 

** (tester:29): CRITICAL **: 05:11:55.610: Exception while loading the model: 

 

aten::_convolution(Tensor input, Tensor weight, Tensor? bias, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups, bool benchmark, bool deterministic, bool cudnn_enabled) -> (Tensor):

Expected at most 12 arguments but found 13 positional arguments.

:

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/conv.py(442): _conv_forward

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/conv.py(446): forward

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1090): _slow_forward

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1102): _call_impl

/usr/src/app/yolov5/models/common.py(49): forward_fuse

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1090): _slow_forward

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1102): _call_impl

/usr/src/app/yolov5/models/common.py(207): forward

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1090): _slow_forward

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1102): _call_impl

/usr/src/app/yolov5/models/yolo.py(149): _forward_once

/usr/src/app/yolov5/models/yolo.py(126): forward

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1090): _slow_forward

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1102): _call_impl

/opt/conda/lib/python3.8/site-packages/torch/jit/_trace.py(958): trace_module

/opt/conda/lib/python3.8/site-packages/torch/jit/_trace.py(741): trace

export.py(71): export_torchscript

export.py(372): run

/opt/conda/lib/python3.8/site-packages/torch/autograd/grad_mode.py(28): decorate_context

export.py(430): main

export.py(435): <module>

Serialized   File "code/__torch__/torch/nn/modules/conv.py", line 12

    bias = self.bias

    weight = self.weight

    input0 = torch._convolution(input, weight, bias, [1, 1], [1, 1], [1, 1], False, [0, 0], 1, False, False, True, True)

             ~~~~~~~~~~~~~~~~~~ <--- HERE

    return input0

 

 

** (tester:29): CRITICAL **: 05:11:55.610: Failed to load model

 

It looks like something that happens when the yolov5 pytorch version and nnstreamer runtime pytorch version are different.

But I heard as pytorch is backward compatible type. 

Yolov5 pytorch version is 1.10.1

And with description on nnstreamer-pytorch in GitHub, It looks like It is built from pytorch version 1.3.1

After this point, I lost the way to go.

 

If do you have any advise, tell me.

 

Additionaly, In tensorfilter element, How can I get the property that name of “outputname” and “inputname” from pytorch model?

When I convert pytorch model to onnx model, there have arg that I can specify.

But I cannot find those arg when I export model as normal pt or tensorsciprt.

 

regards.

 

 


Re: Private: Re: [NNStreamer Technical Discuss] Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

MyungJoo Ham
 

According to the log you've send in the previous message, calling pytorch's load function has failed:

  try {

#ifdef PYTORCH_VER_ATLEAST_1_2_0

    model = std::make_shared<torch::jit::script::Module> (torch::jit::load (model_path));

#else

    model = torch::jit::load (model_path);

#endif

  } catch (const std::invalid_argument &ia) {

    ml_loge ("Invalid argument while loading the model: %s"ia.what ());

    return -1;

  } catch (const std::exception &ex) {

    ml_loge ("Exception while loading the model: %s"ex.what ());

    return -1;

 

the exception is from "Exception while loading the model", which means that the model you've entered

has failed with "load()" function.

Please check if the pytorch invoked by nnstreamer is correct (depending on your own shard-library configurations, different pytorch might be invoked when you call it via nnstreamer/gstreamer). Check if the corresponding pytorch instance is compatible with your model. Then, check if your tensor_filter arguments, "inputname", "outputname" are correct, too.

 

Cheers,

MyungJoo.

ps. Please keep this thread public.
ps.2 If you want more active & efficient discussion, you can try github/issue

 

--------- Original Message ---------

Sender : 이병헌 <byungs286@...>

Date : 2021-12-02 15:51 (GMT+9)

Title : Private: Re: [NNStreamer Technical Discuss] Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

 

Thanks for reply 

 

I've tried your advice as like below

 

  const char *string = "rtspsrc location=rtsp://address:port/mount latency=0 protocols=4 ! \

rtph265depay ! \

avdec_h265 ! \

videoscale ! \

videoconvert ! \

video/x-raw,format=RGB,width=640,height=640 ! \

tensor_converter ! \

tensor_transform mode=typecast option=float32 ! \

tensor_filter framework=pytorch \

model=../../tf_model/yolov5s.torchscript.pt \

input=3:640:640:1 inputname=x inputtype=float32 \

output=1:25200:85 outputname=416 outputtype=float32 ! \

tensor_sink name=tensor_sink";

 

but I still have same error as below..

 

failed to initialize the object: PyTorch

0:00:00.159368517   264 0x564b2683b140 WARN                GST_PADS gstpad.c:1149:gst_pad_set_active:<tensorfilter0:sink> Failed to activate pad

** Message06:37:13.822: gpu = 0, accl = cpu

 

** (tester:264): CRITICAL **: 06:37:13.825: Exception while loading the model: 

 

** (tester:264): CRITICAL **: 06:37:13.825: Failed to load model

 

 

failed to initialize the object: PyTorch

 

Is there have any other problem in that pipeline?

 

 

 


Re: Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

MyungJoo Ham
 

... ! video/x-raw,format=RGB,width=640,height=640 ! tensor_converter ! tensor_filter framework=pytorch model=../../tf_model/yolov5s.torchscript.pt input=3:640:640:1 inputname=x inputtype=float32 output=1:25200:85 outputname=416 outputtype=float32 ...

The actual tensor you've entered for tensor_filter is 3:640:640:1 (uint8) and you've stated that the tensor_filter needs to get 3:640:640:1 (float32). In other words, you are pushing uint8[640][640][3] into float32[640][640][3] array.

You need to transform uint8 stream to float32 stream with tensor_transform before you feed it to tensor_filter.

nnstreamer-pytorch 2.1.0 is tested with pytorch 1.1, 1.3, and 1.6; but this is not related with version compatibility.


Which version of pytorch is available with nnstreamer-pytorch 2.1.0.0-0~202111300837~ubuntu18.04.1?

이병헌
 

Hi dev.
I’m trying to use yolov5s tensorscript.pt file that exported from yolov5 GitHub code with nnstreamer pipeline.
I’m working on develop environment with nnstreamer/tools/docker/ubuntu18.04-run/Dockerfile

Below is my test pipeline.

const char *string = "rtspsrc location=rtsp://address:port/mount latency=0 \
protocols=4 ! rtph265depay ! avdec_h265 ! \
videoscale ! videoconvert ! video/x-raw,format=RGB,width=640,height=640 ! \
tensor_converter ! tensor_filter framework=pytorch \
model=../../tf_model/yolov5s.torchscript.pt \
input=3:640:640:1 inputname=x inputtype=float32 \
output=1:25200:85 outputname=416 outputtype=float32 \
! tensor_sink name=tensor_sink;

But with this pipeline, I got error as below

failed to initialize the object: PyTorch
0:00:00.376511035    29 0x559a1ba3d4c0 WARN                GST_PADS gstpad.c:1149:gst_pad_set_active:<tensorfilter0:sink> Failed to activate pad
** Message: 05:11:55.607: gpu = 0, accl = cpu

** (tester:29): CRITICAL **: 05:11:55.610: Exception while loading the model: 

aten::_convolution(Tensor input, Tensor weight, Tensor? bias, int[] stride, int[] padding, int[] dilation, bool transposed, int[] output_padding, int groups, bool benchmark, bool deterministic, bool cudnn_enabled) -> (Tensor):
Expected at most 12 arguments but found 13 positional arguments.
:
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/conv.py(442): _conv_forward
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/conv.py(446): forward
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1090): _slow_forward
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1102): _call_impl
/usr/src/app/yolov5/models/common.py(49): forward_fuse
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1090): _slow_forward
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1102): _call_impl
/usr/src/app/yolov5/models/common.py(207): forward
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1090): _slow_forward
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1102): _call_impl
/usr/src/app/yolov5/models/yolo.py(149): _forward_once
/usr/src/app/yolov5/models/yolo.py(126): forward
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1090): _slow_forward
/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py(1102): _call_impl
/opt/conda/lib/python3.8/site-packages/torch/jit/_trace.py(958): trace_module
/opt/conda/lib/python3.8/site-packages/torch/jit/_trace.py(741): trace
export.py(71): export_torchscript
export.py(372): run
/opt/conda/lib/python3.8/site-packages/torch/autograd/grad_mode.py(28): decorate_context
export.py(430): main
export.py(435): <module>
Serialized   File "code/__torch__/torch/nn/modules/conv.py", line 12
    bias = self.bias
    weight = self.weight
    input0 = torch._convolution(input, weight, bias, [1, 1], [1, 1], [1, 1], False, [0, 0], 1, False, False, True, True)
             ~~~~~~~~~~~~~~~~~~ <--- HERE
    return input0


** (tester:29): CRITICAL **: 05:11:55.610: Failed to load model

It looks like something that happens when the yolov5 pytorch version and nnstreamer runtime pytorch version are different.
But I heard as pytorch is backward compatible type. 
Yolov5 pytorch version is 1.10.1
And with description on nnstreamer-pytorch in GitHub, It looks like It is built from pytorch version 1.3.1
After this point, I lost the way to go.

If do you have any advise, tell me.

Additionaly, In tensorfilter element, How can I get the property that name of “outputname” and “inputname” from pytorch model?
When I convert pytorch model to onnx model, there have arg that I can specify.
But I cannot find those arg when I export model as normal pt or tensorsciprt.

regards.


Re: What is the version of OpenVINO supported by NNStreamer

MyungJoo Ham
 

On Wed, Nov 10, 2021 at 05:23 PM, jnzhao@... wrote: > > Hi: > Before install the NNStreamer,i already installed OpenVINO R20.03. When i try > to run NNStreamer openvino,can't get any AI inference results.So my questions > is: > 1. Which version is supported by nnstreamer using PPA installing.

The current deployment uses 2019R3 according to https://launchpad.net/~nnstreamer/+archive/ubuntu/ppa and https://review.tizen.org/gerrit/gitweb?p=platform/upstream/dldt.git;a=summary

However, this does not mean that nnstreamer supports 2019R3 only. We just didn't test it with other versions. I believe it will support more recent versions as long as openvino keeps backward compatibility.

  1. How could i do to support Openvino R20.03
  1. install nnstreamer, nnstreamer-openvino, openvino from PPA.
  2. try to upgrade openvino library
  3. test. if OK, stop here.
  4. if failed, recompile nnstreamer-openvino (or the whole nnstreamer) with the updated openvino (and its devel packages) packages.
  5. test. if OK, stop here.
  6. if nnstreamer-openvino has build errors with newer versions of openvino while it is ok with older versions, please report it to github issues (and you may send PR resoling that issue.)
  1. Is Intel NCS2 supported by nnstreamer?

Yes. and it is tested with openvino and https://github.com/nnstreamer/nnstreamer/blob/main/ext/nnstreamer/tensor_filter/tensor_filter_movidius_ncsdk2.c

Jon Thanks.


What is the version of OpenVINO supported by NNStreamer

jnzhao@...
 

Hi: 
  Before install the NNStreamer,i already installed OpenVINO R20.03. When i try to run NNStreamer openvino,can't get any AI inference results.So my questions is:
1. Which version is supported by nnstreamer using PPA installing.
2. How could i do to support Openvino R20.03
3. Is Intel NCS2 supported by nnstreamer?
Jon
Thanks.


Re: nnstreamer v1.7.2 build with meson 0.47.0

MyungJoo Ham
 

No, we do not have official patches for meson v0.47 compatibility.

However, there is a WIP (will be revived if I become available for hobby projects) patch for 0.49 and this might be helpful for your need. I'm not sure if it's compatible with 0.47, but set 0.49 because 0.49 was available in the given environment for this instance.

https://github.com/myungjoo/nnstreamer/commit/c5136e3fc5e83c7a8c8127ed74969738e2f56cca#diff-30d8f6be6320feeacf686be94f48c70869b52630e01ea625f0f15adc0d57c3e4

This branch might be removed or updated soon. So if you need it, please fetch it.


nnstreamer v1.7.2 build with meson 0.47.0

naveenkumar16071996.nk@...
 

We understand that to build nnstreamer v1.7.2 requires meson version >=0.50.0. But due to some platform dependency we are trying to build nnstreamer v1.7.2 with yocto meson version 0.47.0. Is it possible to build nnstreamer v1.7.2 with meson v0.47.0 ?

Do we have any patch file available for this?


Re: tensorflow uint8 object detection model

MyungJoo Ham
 

On Thu, Aug 12, 2021 at 01:41 PM, shortcipher wrote:

Thanks for the quick response!  I tried that modification adding the threshold and permuting the output tensor indices and I'm still getting the same error.  (I also tried all other permutations of the indices 0:1:2:3,50 / 3:2:1:0,50 / 3:1:2:0 / etc) Did you make any other changes to get it to work well?

Here is the gst pipeline description that I've tested:

gst-launch-1.0 v4l2src name=cam_src ! videoconvert ! videoscale ! video/x-raw,width=640,height=480,format=RGB,framerate=30/1 ! tee name=t t. ! queue leaky=2 max-size-buffers=2 ! videoscale ! video/x-raw,width=320,height=320,format=RGB ! tensor_converter ! tensor_transform mode=arithmetic option=typecast:uint8,add:0,div:1 ! tensor_filter framework=tensorflow2-lite model=model.tflite ! tensor_decoder mode=bounding_boxes option1=mobilenet-ssd-postprocess option2=coco-labels-paper.txt option3=0:1:2:3,50 option4=640:480 option5=320:320 ! compositor name=mix sink_0::zorder=2 sink_1::zorder=1 ! videoconvert ! ximagesink t. ! queue leaky=2 max-size-buffers=10 ! videoconvert ! mix.

It appears that the video format (RGB) from t might be not compatible with compositor: https://gstreamer.freedesktop.org/documentation/compositor/index.html?gi-language=c


Re: tensorflow uint8 object detection model

shortcipher
 

Thanks for the quick response!  I tried that modification adding the threshold and permuting the output tensor indices and I'm still getting the same error.  (I also tried all other permutations of the indices 0:1:2:3,50 / 3:2:1:0,50 / 3:1:2:0 / etc) Did you make any other changes to get it to work well?

1 - 20 of 62