Add yolov8 ncnn example

[whyb]

Add yolov8 ncnn example

ReadMe
Convert yolov8 model to ncnn model workflow:

step 1:

If you don't want to train the model yourself. You should go to the ultralytics website download the pretrained model file.
original pretrained model from https://docs.ultralytics.com/models/yolov8/#supported-tasks-and-modes

step 2:

run this command.

conda create --name yolov8 python=3.11
conda activate yolov8
pip install ultralytics onnx numpy protobuf

step 3:

save source code file(export_model_to_ncnn.py):

from ultralytics import YOLO
detection_models = [
    ["./Detection-pt/yolov8n.pt", "./Detection-pt/"],
    ["./Detection-pt/yolov8s.pt", "./Detection-pt/"],
    ["./Detection-pt/yolov8m.pt", "./Detection-pt/"],
    ["./Detection-pt/yolov8l.pt", "./Detection-pt/"],
    ["./Detection-pt/yolov8x.pt", "./Detection-pt/"]
]
for model_dict in detection_models:
    model = YOLO(model_dict[0])  # load an official pretrained weight model
    model.export(format="ncnn", dynamic=True, save_dir=model_dict[1], simplify=True)

step 4:

run command:

python export_model_to_ncnn.py

Test Image:

input image(640x640):

output image(640x640):

[tencent-adm]

Thank you for your submission, we really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
1 out of 2 committers have signed the CLA.

✅ nihui
❌ whyb
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[nihui]

Thanks for your contribution !

[SIXGODLI]

通过转换之后，param也只有一个输出，无法使用ncnn自带的yolov8进行推理

[whyb]

通过转换之后，param也只有一个输出，无法使用ncnn自带的yolov8进行推理

yolov8 是一个输入1个，输出1个，您遇到的转换问题是因为没安装好环境，或者您是通过的ultralytics源码安装的，且并未使用requirement.txt，或者是python版本过低。

能否提供一些更加具体的问题描述呢？

如果ultralytics官方转换ncnn的方式实在走不通，不妨试试换一个方向：导出torchscript 格式的模型，然后再使用pnnx转换到ncnn模型。

[git-zhan]

按照examples/yolov8.cpp 教程转换成ncnn后用C++推理报错

[whyb]

按照examples/yolov8.cpp 教程转换成ncnn后用C++推理报错

您截图上的报错是因为没有读取到对应的模型文件：yolov8n.param 和 yolov8n.bin ，可能的原因很多，有可能因为权限，有可能因为路径包含亚洲字符，也可能模型文件不存在。