Files
FastDeploy/examples/vision/detection/scaledyolov4
huangjianhui 376fdbfe2c [Other] Update old Api to new ones (#861)
* Update keypointdetection result docs

* Update im.copy() to im in examples

* Update new Api, fastdeploy::vision::Visualize to fastdeploy::vision

* Update SwapBackgroundSegmentation && SwapBackgroundMatting to SwapBackground

* Update README_CN.md

* Update README_CN.md
2022-12-14 17:25:58 +08:00
..

ScaledYOLOv4准备部署模型

导出ONNX模型

访问ScaledYOLOv4官方github库按照指引下载安装下载scaledyolov4.pt 模型,利用 models/export.py 得到onnx格式文件。如果您导出的onnx模型出现问题,可以参考ScaledYOLOv4#401的解决办法

#下载ScaledYOLOv4模型文件
Download from the goole drive https://drive.google.com/file/d/1aXZZE999sHMP1gev60XhNChtHPRMH3Fz/view?usp=sharing

# 导出onnx格式文件
python models/export.py  --weights PATH/TO/scaledyolov4-xx.pt --img-size 640

下载预训练ONNX模型

为了方便开发者的测试下面提供了ScaledYOLOv4导出的各系列模型开发者可直接下载使用。下表中模型的精度来源于源官方库

模型 大小 精度
ScaledYOLOv4-P5-896 271MB 51.2%
ScaledYOLOv4-P5+BoF-896 271MB 51.7%
ScaledYOLOv4-P6-1280 487MB 53.9%
ScaledYOLOv4-P6+BoF-1280 487MB 54.4%
ScaledYOLOv4-P7-1536 1.1GB 55.0%
ScaledYOLOv4-P5 271MB -
ScaledYOLOv4-P5+BoF 271MB -
ScaledYOLOv4-P6 487MB -
ScaledYOLOv4-P6+BoF 487MB -
ScaledYOLOv4-P7 1.1GB -

详细部署文档

版本说明