bharath5673/deepstream

yolov5, yolov8, segmenations, face, pose, keypoints on deepstream

Deepstream deepstream-6.3-ubuntu20.04

1. Make sure that you install Deepstream 6.3 as shown in this Guide

2. Clone the repository recursively:

   git clone --recurse-submodules  https://github.com/bharath5673/Deepstream.git
   

   If you already cloned and forgot to use --recurse-submodules you can run git submodule update --init

3. follow the instructions as shown in the repos..

---

---

DeepStream MultiModel

@ https://github.com/bharath5673/Deepstream/tree/main/DeepStream-Configs/DeepStream-MultiModel

---

---

ROI based counts on deepstream

@ https://github.com/bharath5673/Deepstream/tree/main/DeepStream-Python

---

---

Trajectory tracking on deepstream

@ https://github.com/bharath5673/Deepstream/tree/main/DeepStream-Python

---

---

Custom CNN to DeepStream in simple 3 steps

START @ https://github.com/bharath5673/Deepstream/tree/main/CNN-to-DeepStream

---

---

Deepstream ONNXified Quick DEMO

cd Deepstream
sh QuickDemo.sh

---

---

Deepstream multistreams, tiled, multimodels, customizations and extra configs

@ https://github.com/bharath5673/Deepstream/tree/main/DeepStream-Configs/test

---

---

Deepstream Multimodels python

@https://github.com/bharath5673/Deepstream/blob/main/DeepStream-Python/deepstream_test_yolo_track_multimodel.py

---

---

Easy steps installation

@ https://gist.github.com/bharath5673/800a18cc7474ce9c22fda6deaaa98354

---

---

Acknowledgements

Expand

• https://github.com/marcoslucianops/DeepStream-Yolo
• https://github.com/AlexeyAB/darknet
• https://github.com/WongKinYiu/yolor
• https://github.com/WongKinYiu/PyTorch_YOLOv4
• https://github.com/WongKinYiu/ScaledYOLOv4
• https://github.com/Megvii-BaseDetection/YOLOX
• https://github.com/ultralytics/yolov3
• https://github.com/ultralytics/yolov5
• https://github.com/DingXiaoH/RepVGG
• https://github.com/JUGGHM/OREPA_CVPR2022
• https://github.com/TexasInstruments/edgeai-yolov5/tree/yolo-pose
• https://github.com/ultralytics/ultralytics

deepstream's Issues

YOLOX Fps difference between deepstream-app and python script

Hello, Thank you for the amazing repository!

I managed to setup the repository and the requirements thanks to the amazing gist guide! Thank you.

I could run the python script for yolox model with python3 deepstream-yolo-track.py <vid_dir>

However, there seems to be a difference between the FPS between the two processes.

First process with deepstream-app < config_file > managed to get around 100 FPS ( with yolox_small model )
Second process with python3 deepstream-yolo-track.py managed to get around 30 FPS ( with yolox_small model )

Do I need to change any configurations to get the same FPS as the deepstream-app or is it just due to the python binding?

INT8 calibration

Hi,
How can we get a Tensor RT file (.engine) exported with int8 precision? It is important as with fp16 I get too low FPS. Thanks in advance.

Can you open source libnvdsinfer_custom_impl_Yolo.so ? Thanks

Or NvDsInferParseCustomYolo func. How it works ?

Do you have a plan to implement instance segmentation of yolov5

Do you have a plan to implement instance segmentation of yolov5, thanks.

thrust::system::detail::bad_alloc

I make .so successful , but got this error while run deepstream_test_1 :

Playing file /opt/nvidia/deepstream/deepstream/samples/streams/sample_720p.h264
Adding elements to Pipeline

Linking elements in the Pipeline

Starting pipeline

terminate called after throwing an instance of 'thrust::system::detail::bad_alloc'
what(): std::bad_alloc: cudaErrorUnsupportedPtxVersion: the provided PTX was compiled with an unsupported toolchain.
Aborted (core dumped)

ModuleNotFoundError: No module named ‘common’

blockIdx blockDim threadIdx was not declared in this scope

yolo-Tracker for mp4 and RTSP

Hi,

Firstly, thank you very much for this powerful repository with concise step-by-step instructions.
All the commands run flawlessly and I am able to implement Yolov5 detection on Jetson Xavier. The FPS is amazing.

I am also able to implement the yolo-Tracker.py given is Step 11 of https://gist.github.com/bharath5673/800a18cc7474ce9c22fda6deaaa98354/

I would like to put forth one gentle request. The yolo-Tracker requires .h264 as mentioned in Step 11.
Can I also use it directly on an RTSP or .mp4?
If yes, please suggest the configuration changes that I need to do to achieve the same.
When I tried the mp4 video file, it threw an error and exited the program.

Thank you once again!

Unable to create NvStreamMux

I just install everthying as discribed on paper (tensorRT, deepstream 6.1, cuda and cudnn libs) but this error comes:

Unable to create NvStreamMux
Traceback (most recent call last):
File "deepstream-yolo.py", line 538, in
sys.exit(main(sys.argv))
File "deepstream-yolo.py", line 405, in main
pipeline.add(streammux)
TypeError: Argument 1 does not allow None as a value

Step 10: Running yolo.py Error

thanks your great job. i followed your step(yolo5n) and get some problem.

from step1 to step9. everything is great.

step8 deepstream-app -c deepstream_app_config.txt

**PERF: FPS 0 (Avg)
**PERF: 0.00 (0.00)
** INFO: <bus_callback:194>: Pipeline ready

** INFO: <bus_callback:180>: Pipeline running

**PERF: 260.93 (260.91)
** INFO: <bus_callback:217>: Received EOS. Exiting ...

Quitting
App run successful.

Step 10: Running yolo.py

root@3c38dad9654b:/opt/nvidia/deepstream/deepstream-6.1/DeepStream-Yolo# python3 deepstream-yolo.py file:///opt/nvidia/deepstream/deepstream/samples/streams/sample_1080p_h264.mp4
deepstream-yolo.py:387: PyGIDeprecationWarning: Since version 3.11, calling threads_init is no longer needed. See: https://wiki.gnome.org/PyGObject/Threading
GObject.threads_init()
Creating Pipeline

Creating streamux

Creating source_bin 0

Creating source bin
source-bin-00
Creating Pgie

Creating tiler

Creating nvvidconv

Creating nvosd

Creating EGLSink

Adding elements to Pipeline

Linking elements in the Pipeline

deepstream-yolo.py:514: PyGIDeprecationWarning: GObject.MainLoop is deprecated; use GLib.MainLoop instead
loop = GObject.MainLoop()
Now playing...
1 : file:///opt/nvidia/deepstream/deepstream/samples/streams/sample_1080p_h264.mp4
Starting pipeline

WARNING: [TRT]: CUDA lazy loading is not enabled. Enabling it can significantly reduce device memory usage. See CUDA_MODULE_LOADING in https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars
Deserialize yoloLayer plugin: yolo
WARNING: [TRT]: TensorRT was linked against cuDNN 8.6.0 but loaded cuDNN 8.4.0
WARNING: [TRT]: TensorRT was linked against cuDNN 8.6.0 but loaded cuDNN 8.4.0
WARNING: [TRT]: CUDA lazy loading is not enabled. Enabling it can significantly reduce device memory usage. See CUDA_MODULE_LOADING in https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars
0:00:05.177224055 21150 0x3209060 INFO nvinfer gstnvinfer.cpp:646:gst_nvinfer_logger: NvDsInferContext[UID 1]: Info from NvDsInferContextImpl::deserializeEngineAndBackend() <nvdsinfer_context_impl.cpp:1900> [UID = 1]: deserialized trt engine from :/opt/nvidia/deepstream/deepstream-6.1/DeepStream-Yolo/model_b1_gpu0_fp32.engine
INFO: ../nvdsinfer/nvdsinfer_model_builder.cpp:610 [Implicit Engine Info]: layers num: 5
0 INPUT kFLOAT data 3x640x640
1 OUTPUT kFLOAT num_detections 1
2 OUTPUT kFLOAT detection_boxes 25200x4
3 OUTPUT kFLOAT detection_scores 25200
4 OUTPUT kFLOAT detection_classes 25200

0:00:05.236333287 21150 0x3209060 INFO nvinfer gstnvinfer.cpp:646:gst_nvinfer_logger: NvDsInferContext[UID 1]: Info from NvDsInferContextImpl::generateBackendContext() <nvdsinfer_context_impl.cpp:2003> [UID = 1]: Use deserialized engine model: /opt/nvidia/deepstream/deepstream-6.1/DeepStream-Yolo/model_b1_gpu0_fp32.engine
0:00:05.601797793 21150 0x3209060 INFO nvinfer gstnvinfer_impl.cpp:328:notifyLoadModelStatus: [UID 1]: Load new model:config_infer_primary_yoloV5.txt sucessfully
Decodebin child added: source

Decodebin child added: decodebin0

Decodebin child added: qtdemux0

Decodebin child added: multiqueue0

Decodebin child added: h264parse0

Decodebin child added: capsfilter0

Decodebin child added: aacparse0

Decodebin child added: avdec_aac0

Decodebin child added: nvv4l2decoder0

In cb_newpad

gstname= video/x-raw
features= <Gst.CapsFeatures object at 0x7ff4c2cb77c0 (GstCapsFeatures at 0x7ff3bc00acc0)>
In cb_newpad

gstname= audio/x-raw
0:00:06.783651849 21150 0x18e4d80 WARN nvinfer gstnvinfer.cpp:2299:gst_nvinfer_output_loop: error: Internal data stream error.
0:00:06.783783685 21150 0x18e4d80 WARN nvinfer gstnvinfer.cpp:2299:gst_nvinfer_output_loop: error: streaming stopped, reason not-negotiated (-4)
Error: gst-stream-error-quark: Internal data stream error. (1): gstnvinfer.cpp(2299): gst_nvinfer_output_loop (): /GstPipeline:pipeline0/GstNvInfer:primary-inference:
streaming stopped, reason not-negotiated (-4)
Exiting app

Thank you for any kind of help. I will appreciate very much.

Floating point exception (core dumped)

Getting error while running deepstream-app -c deepstream_app_config_yolo.txt

Other repository is working fine.
Can you please help here?