Android 通过ncnn加载darknet模型

1. opencv环境部署
   参考文件Android Studio下opencv部署
   注意，在这里实际不需要第3步-导入opencv包，直接把include头文件和.so动态库放进对应的目录就好，其他的按博主示例操作。
2. ncnn部署
   参考文章Android Studio下ncnn部署
   基本是一个套路，下载第三方的包，导入头文件和库，配置CMake
   3.模型转换
   参考文章ncnn模型转换

你下载的ncnn里面就有，window环境下模型转换操作如下：

cd C:\Users\login\Desktop\ncnn-20230816-windows-vs2022\x64\bin

C:\Users\login\Desktop\ncnn-20230816-windows-vs2022\x64\bin>onnx2ncnn.exe
Usage: onnx2ncnn.exe [onnxpb] [ncnnparam] [ncnnbin]

C:\Users\login\Desktop\ncnn-20230816-windows-vs2022\x64\bin>dir
 驱动器 C 中的卷是 Windows
 卷的序列号是 96B3-D72F

 C:\Users\login\Desktop\ncnn-20230816-windows-vs2022\x64\bin 的目录

2023/08/16  12:54    <DIR>          .
2023/08/16  12:54    <DIR>          ..
2023/08/16  12:41         1,153,024 caffe2ncnn.exe
2023/08/16  12:41            81,920 darknet2ncnn.exe
2023/08/16  12:41            86,528 mxnet2ncnn.exe
2023/08/16  12:50         9,584,640 ncnn2int8.exe
2023/08/16  12:50         9,422,336 ncnn2mem.exe
2023/08/16  12:50         9,690,624 ncnn2table.exe
2023/08/16  12:41            17,920 ncnnmerge.exe
2023/08/16  12:50         9,662,464 ncnnoptimize.exe
2023/08/16  12:41           984,576 onnx2ncnn.exe
               9 个文件     40,684,032 字节
               2 个目录 148,911,742,976 可用字节

C:\Users\login\Desktop\ncnn-20230816-windows-vs2022\x64\bin>darknet2ncnn.exe
Usage: darknet2ncnn.exe [darknetcfg] [darknetweights] [ncnnparam] [ncnnbin] [merge_output]
        [darknetcfg]     .cfg file of input darknet model.
        [darknetweights] .weights file of input darknet model.
        [cnnparam]       .param file of output ncnn model.
        [ncnnbin]        .bin file of output ncnn model.
        [merge_output]   merge all output yolo layers into one, enabled by default.

C:\Users\login\Desktop\ncnn-20230816-windows-vs2022\x64\bin>darknet2ncnn.exe  D:\Native\app\src\main\assets\model.cfg D:\Native\app\src\main\assets\model.weights D:\Native\app\src\main\assets\model.param D:\Native\app\src\main\assets\model.bin
Loading cfg...
WARNING: The ignore_thresh=0.700000 of yolo0 is too high. An alternative value 0.25 is written instead.
WARNING: The ignore_thresh=0.700000 of yolo1 is too high. An alternative value 0.25 is written instead.
Loading weights...
Converting model...
194 layers, 211 blobs generated.
NOTE: The input of darknet uses: mean_vals=0 and norm_vals=1/255.f.
NOTE: Remember to use ncnnoptimize for better performance.

C:\Users\login\Desktop\ncnn-20230816-windows-vs2022\x64\bin>