hpc203/bytetrack-opencv-onnxruntime

分别使用OpenCV、ONNXRuntime部署YOLOX+ByteTrack目标跟踪，包含C++和Python两个版本的程序

bytetrack-opencv-onnxruntime

使用OpenCV部署YOLOX+ByteTrack目标跟踪，包含C++和Python两个版本的程序。

使用ONNXRuntime部署YOLOX+ByteTrack目标跟踪，包含C++和Python两个版本的程序。

由于.onnx文件超过25M，无法直接上传。.onnx文件和测试文件sample.mp4 在百度云盘，链接: https://pan.baidu.com/s/1dYz_Ru5EgFg_DZhdDHH-Sg 密码: iwtm

安装eigen库和编译C++程序的指令，详情请看文件夹里的readme

运行python程序之前，需要先安装依赖库。如果在安装cython_bbox失败， 请参见文章 https://www.jianshu.com/p/51f7f7f8b262

bytetrack-opencv-onnxruntime's Issues

Tracking Outputs vs Detection Outputs

While using ByteTrack as a tracker, the visualized bounding boxes sometimes do not precisely match the detection results. Are there any methods to assign tracking IDs to detections without altering the bounding box coordinates? I am asking this question because the tracking bounding boxes are sometimes too wide for my specific case.

[Onnx 读取错误]

您好！ 使用您的代码和提供的onnx文件会提示读取错误: 主要是slice问题，该怎么解决呢？
Node [Slice]:(467) parse error: OpenCV(4.5.1)

请问是支持GPU加速的吗

atlbrs.push_back(atracks[i]->tlbr);这个值应该是kalman predict后，更新后的值了吧？

您好，首先感谢你的分享。
utils.cpp文件的BYTETracker::iou_distance()函数中。
在IOU匹配之前，要取轨迹预测的位置用来和当前detection进行匹配。
但是我看了一下代码流程，
atlbrs.push_back(atracks[i]->tlbr);
atracks[i]->tlbr的值不是predict之后的值吧？我看了下，好像没更新。
我想问下，这个地方我理解的是不是有问题。

怎么重置序号

我在c++的实际使用中想要重置一下序号，让它回到0重新开始计数，请问改怎么修改

Run in opencv-cpp.

Total frames: 1625, fps: 29
terminate called after throwing an instance of 'cv::Exception'
what(): OpenCV(4.2.0) /home/dk/opencv/modules/dnn/src/onnx/onnx_importer.cpp:101: error: (-211:One of the arguments' values is out of range) Input is out of OpenCV 32S range in function 'convertInt64ToInt32'

I have encountered the above problems and hope to get your help.

多类别

多类别的请问有相应的工程嘛

some problems about high_thresh

the code in bytetrack-opencv-onnxruntime/onnxruntime/cpp/src/BYTETracker.cpp script
if (track->score < this->high_thresh)
continue;
low score dets in the first frame will not use. is it right?
if so, is it necessary to Add a restriction, such as track->frame_id
Maybe I misunderstood it. Hope your reply.
have a nice day.

读取onnx模型报错

[ERROR:0] global /home/opencv/modules/dnn/src/onnx/onnx_importer.cpp (1788) handleNode DNN/ONNX: ERROR during processing node with 5 inputs and 1 outputs: [Slice]:(467)
terminate called after throwing an instance of 'cv::Exception'
what(): OpenCV(4.5.0) /home/opencv/modules/dnn/src/onnx/onnx_importer.cpp:1797: error: (-2:Unspecified error) in function 'handleNode'

Node [Slice]:(467) parse error: OpenCV(4.5.0) /home/opencv/modules/dnn/src/onnx/onnx_importer.cpp:697: error: (-2:Unspecified error) in function 'void cv::dnn::dnn4_v20200908::ONNXImporter::handleNode(const opencv_onnx::NodeProto&)'

Slice layer only supports steps = 1 (expected: 'countNonZero(step_blob != 1) == 0'), where
'countNonZero(step_blob != 1)' is 1
must be equal to
'0' is 0

Hello, I have encountered the above problems and hope to get your help. Thanks!

请问如果要换Yolov5的话，需要改动哪些

关于C++下使用opencv部署模型出现的问题

非常出色的bytetrack的cpu部署版本。感谢您的分享。已经在C++下成功使用ONNXRuntime部署ByteTrack。
但是关于opencv的部署在模型载入，即readnet时出现了问题
opencv版本为4.5，出现bug如下

Total frames: 1625, fps: 29
start load model
[ERROR:0] global /home/chiye/software/opencv-4.5.0/modules/dnn/src/onnx/onnx_importer.cpp (1786) handleNode DNN/ONNX: ERROR during processing node with 5 inputs and 1 outputs: [Slice]:(467)
terminate called after throwing an instance of 'cv::Exception'
  what():  OpenCV(4.5.0) /home/chiye/software/opencv-4.5.0/modules/dnn/src/onnx/onnx_importer.cpp:1797: error: (-2:Unspecified error) in function 'handleNode'
> Node [Slice]:(467) parse error: OpenCV(4.5.0) /home/chiye/software/opencv-4.5.0/modules/dnn/src/onnx/onnx_importer.cpp:697: error: (-2:Unspecified error) in function 'void cv::dnn::dnn4_v20200908::ONNXImporter::handleNode(const opencv_onnx::NodeProto&)'
> > Slice layer only supports steps = 1 (expected: 'countNonZero(step_blob != 1) == 0'), where
> >     'countNonZero(step_blob != 1)' is 1
> > must be equal to
> >     '0' is 0
> 
Aborted (core dumped)

请问这是哪里出现了问题？期待您的早日回复

关于yolox的构造函数

up主你好，我看到你代码里有几行不是特别明白，yolox::yolox()
string model_path = "/home/ByteTrack/byte_tracker/model/bytetrack_s.onnx";
为啥YOLOx的构造函数里面读取的是bytetrack_s.onnx的模型？其实我可以理解为是对跟踪模型的读取，请问放在这里面有特别的含义么？