/NCNN_ObjectDetection

MobileNet SSD/YOLOv4/YOLOv5/YOLOX etc Obejct Detection on Android Camera

Primary LanguageC++MIT LicenseMIT

Android Camera Object Detection

Author: IronSublimate

Source Code: https://github.com/IronSublimate/NCNN_ObjectDetection

use Tencent's NCNN framework to run yolov4-tiny yolov5 and mobilenet-ssd
NCNN Verison : 20220420

How to Build and Run

step1

https://github.com/Tencent/ncnn/releases

download ncnn-android-vulkan-lib.zip or build ncnn for android yourself

step2

extract ncnn-android-vulkan-lib.zip into app/src/main/jni or change the ncnn path to yours in app/src/main/jni/CMakeLists.txt

step3

open this project with Android Studio, build it and enjoy!

screenshot

Development

  1. Write a java class extends NCNNDetector and fininsh 3 method inherited from NCNNDetector. These methods might be native methods.

    public boolean Init(AssetManager mgr);
public NCNNDetector.Obj[] Detect(Bitmap bitmap, boolean use_gpu);
public boolean Deinit();

  2. Write *jni.cpp to define native methods above and add your cpp in CMakeLists.txt.

    add_library(
        ncnn_detector SHARED
        NCNN_detector_public_jni.cpp
        mobilenetssdncnn_jni.cpp
        yolov5ncnn_jni.cpp
        # add your cpp here
)

    Warning If there are more than two dynamic library(.so) both linking to libncnn.a, you will get Fatal signal 6 (SIGABRT) in running fuction load_param. You must compile all *jni.cpp into one dynamic library(libncnn_detector.so).

    See also Tencent/ncnn#976

  3. Add your class in MainActivity.java

    public class MainActivity extends AppCompatActivity {
    //...
    private static final HashMap<String, String> detectNetwork = new HashMap<String, String>() {{
        // class name ： method name
        //method name shows in GUI
        //class name is used to reflect
        put(MobilenetSSDNcnn.class.getName(), "MobileNet SSD");
        put(YoloV5Ncnn.class.getName(), "YOLOv5");
        put(NanoDet.class.getName(), "Nano Det");
    put(YOLOv4Tiny.class.getName(),"YOLOv4 Tiny");
    }};
    //...
}

Known Issues

  1. Might be crashed when first open this app
  2. only ncnn version 1.0.20220420 and before is supported, other version will crash in yolox_s-int8 
         yoloXInt8.load_model(mgr, "yolox_s-int8.bin");

Acknowledgement