YOLOv8-TensorRT

YOLOv8-TensorRT copied to clipboard

cv::dnn::blobFromImage(image, nchw, 1 / 255.f, size, cv::Scalar(0, 0, 0), true, false, CV_32F);函数耗时

7

使用后推理时间大幅提升，但是这行代码耗时最久，cv::dnn::blobFromImage(image, nchw, 1 / 255.f, size, cv::Scalar(0, 0, 0), true, false, CV_32F)。

whiteCupz

pose_postprocess for multiple classes

3

I trained the pose model with 2 classes. Each class has 5 keypoints. I successfully converted the model, but I'm unable to extract the multiple classes from the output. How...

saim212

terminate called after throwing an instance of 'std::logic_error'

2

I've got an error: ``` terminate called after throwing an instance of 'std::logic_error' what(): basic_string::_M_construct null not valid Aborted (core dumped) ``` I've built successfully the whole C++ project. When...

ablyzniuk

“unistd.h”: No such file or directory

8

have u guys meet this problem?just after cmake, trying to build exe

Chlhacker

it is very terrible running on Windows

9

Dwinovo

Input size variety.

1

How to keep different picture sizes(I meant dynamic input shape size, not only alongside batch axis) as the original ultralytics approach describes?

ablyzniuk

Static Symbolic Method Operator

1

你好，在 TRT_NMS 的 symbolic 函数中使用了 `TRT::EfficientNMS_TRT`，这看起来是使用了 [efficientNMSPlugin](https://github.com/NVIDIA/TensorRT/tree/main/plugin/efficientNMSPlugin)，但这里使用的参数与官方文档中有所不同，添加了代表数据类型的后缀，这是为什么？另外我想知道 TensorRT 中的 Plugin 是如何支持直接在 g.op() 中使用的，我查看了 PyTorch 的官方文档，里面仅给出了 [TorchScript operators](https://pytorch.org/docs/2.2/onnx_torchscript_supported_aten_ops.html) 和 [ONNX standard operator](https://github.com/onnx/onnx/blob/master/docs/Operators.md) ``` out = g.op('TRT::EfficientNMS_TRT', boxes, scores, iou_threshold_f=iou_threshold, score_threshold_f=score_threshold, max_output_boxes_i=max_output_boxes,...

fy-shen

YOLOv8-TensorRT and ByteTrack

2

Hi @triple-Mu, I wanted to check in about @pax7's request from issue #85. Could you please commit your example that uses YOLOv8-TensorRT and ByteTrack in detection to the repository? Thanks!

dahan198

Conversion steps for INT8

3

Hello, I'm trying to do inference with INT8 instead of F32. But the steps in the sample README.md haven't done much help. How to do this conversion? i have tried...

suryan-s

export engine error

1

python build.py --weights yolov8n.onnx --iou-thres 0.65 --conf-thres 0.25 --topk 100 --fp16 --device cuda:0 I tried to run build.py to export engine. However, i encountered an error 

quxianjiuguo