Skip to content

使用darknet训练yolov4-tiny网络模型以进行人脸口罩的目标检测任务,模型由yolo格式转换为ncnn格式并部署于android

License

Notifications You must be signed in to change notification settings

SunnyGrocery/ncnn-android-yolov4-tiny-custom

Repository files navigation

ncnn-android-yolov4-tiny-custom

Face Mask Detection

Model trained by AlexeyAB/darknet,which implement the YOLOv4-tiny algorithm, Model converted and deployed by Tencent/NCNN.

Introduction

  • APP对人脸是否佩戴口罩进行目标检测;
  • 模型在darknet上训练;
  • 数据集包含12,373 张图片,训练集与测试集比例为8:2;
  • 使用yolov4-tiny,并在yolov4-tiny.conv.29上进行预训练;
  • 通过NCNN完成模型转换与安卓平台部署,可在Android端调用摄像头实时检测,在骁龙865设备上可达20FPS。

Build

  • Using Android Studio to build this project.
  • You need install Android SDK , Android NDK and CMake additional in Android Studio.

Enjoy it!

  • releases (In order to reduce .apk size, apk retain the environment of arm64 only.)

Example(Original Version)

img


Reference:

About

使用darknet训练yolov4-tiny网络模型以进行人脸口罩的目标检测任务,模型由yolo格式转换为ncnn格式并部署于android

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published