NvDeepSort People Tracking For All Cameras

Accelerated Computing Intelligent Video Analytics DeepStream SDK
,
1

**• Hardware Platform : Ubuntu 20.04 x64 RTX4090
**• DeepStream Version: 6.3
**• TensorRT Version: Tao toolkit 5.2.0
**• NVIDIA GPU Driver Version : 530.41.03
**• Issue Type : Question

Hello,

I am trying to run NvDeepSort on DS 6.3, my aim is to implement tracking through the PeopleNet model.

I can track it with NvSORT, but when people leave the camera, their IDs change.

I think more precise tracking can be done with NvDeepSort+Re-ID.

When I download the Re-ID model from this link, only the onnx or etlt file is available.

Deepsort configuration file expects the following files from me.

tltEncodedModel: “/opt/nvidia/deepstream/deepstream/samples/models/Tracker/resnet50_market1501.etlt”
tltModelKey: “nvidia_tao”
modelEngineFile: “/opt/nvidia/deepstream/deepstream/samples/models/Tracker/resnet50_market1501.etlt_b100_gpu0_fp16.engine”

I do not have a file with .engine extension.

Normally, I can create an engine file this way, but it requires an “int8 calibration file” and this file does not exist.

tao-converter /path/resnet50_market1501.etlt
-k nvidia_tlt
-c /path/int8_calibration.txt
-o output_bbox/BiasAdd,output_cov/Sigmoid
-d 3,416,736
-i nchw
-m 1
-t int8
-e /path/model.engine

When I run DS6.3 without the .engine file, it says the .engine file is being created.

But file not created and it gives an error.

Element: nvtracker, Lib File: libnvds_nvmultiobjecttracker.so

gstnvtracker: Loading low-level lib at /opt/nvidia/deepstream/deepstream/lib/libnvds_nvmultiobjecttracker.so
~~ CLOG[src/modules/ReID/ReID.cpp, loadTRTEngine() @line 583]: Engine file does not exist
[NvMultiObjectTracker] Load engine failed. Create engine again.
WARNING: [TRT]: CUDA lazy loading is not enabled. Enabling it can significantly reduce device memory usage and speed up TensorRT initialization. See “Lazy Loading” section of CUDA documentation CUDA C++ Programming Guide
WARNING: [TRT]: CUDA lazy loading is not enabled. Enabling it can significantly reduce device memory usage and speed up TensorRT initialization. See “Lazy Loading” section of CUDA documentation CUDA C++ Programming Guide
Process finished with exit code 139 (interrupted by signal 11:SIGSEGV)

I need is to test it on RTX4090 first and then run it on Orin Nano.

Thanks.

3

Can you setup model based on: Gst-nvtracker — DeepStream documentation 6.4 documentation

4

Hello, I am keeping this issue waiting, first of all I am trying to solve the following problem.

Thanks.

5

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks

The topic in Orin Nano is closed already. Kindly let us know if you still need support here, otherwise we’ll close this topic, thank you.