Still cannot run custom training model in this Project.

[hoanhvmetavi]

I, i have an issue about using my trainning model.

I am following this issue in:
#1 #3 #8

But when running app in android studio, the app crashed. Change back your model in to the app, it worked nomarlly.
My onnx model can run infer normal, so i think that my ncnn model has an issue or the app does not run my model correctly.
I am stuck at this for some days, so please give me a hand.

Many thanks.

[hoanhvmetavi]

Logs:

03/10 10:18:58: Launching 'app' on Pixel XL API 30.
Install successfully finished in 273 ms.
$ adb shell am start -n "com.tencent.ncnnyolox/com.tencent.ncnnyolox.MainActivity" -a android.intent.action.MAIN -c android.intent.category.LAUNCHER
Connected to process 9990 on device 'Pixel_XL_API_30 [emulator-5554]'.
Capturing and displaying logcat messages from application. This behavior can be disabled in the "Logcat output" section of the "Debugger" settings page.
I/ncent.ncnnyolo: Late-enabling -Xcheck:jni
I/ncent.ncnnyolo: Unquickening 12 vdex files!
W/ncent.ncnnyolo: Unexpected CPU variant for X86 using defaults: x86
D/ApplicationLoaders: Returning zygote-cached class loader: /system/framework/android.test.base.jar
I/ncent.ncnnyolo: The ClassLoaderContext is a special shared library.
D/NetworkSecurityConfig: No Network Security Config specified, using platform default
D/NetworkSecurityConfig: No Network Security Config specified, using platform default
D/libEGL: loaded /vendor/lib/egl/libEGL_emulation.so
D/libEGL: loaded /vendor/lib/egl/libGLESv1_CM_emulation.so
D/ncnn: JNI_OnLoad
D/libEGL: loaded /vendor/lib/egl/libGLESv2_emulation.so
D/ncnn: loadModel 0xe76c29c0
D/ncnn: openCamera 0
W/NdkCamera: open
W/NdkCamera: open 1 270
E/ACameraManager: openCamera: connect camera device failed: Status(-8, EX_SERVICE_SPECIFIC): '1: validateClientPermissionsLocked:1165: Caller "" (PID 10123, UID 9990) cannot open camera "1" without camera permission'
E/NdkCameraDevice: ACameraDevice_createCaptureRequest: invalid argument! device 0x0 request 0xe7d12f90
E/NdkCaptureRequest: ACaptureRequest_addTarget: Error: invalid input: req 0x0, req-targets 0x0, target 0xe7e7a8d0
E/NdkCameraDevice: ACameraDevice_createCaptureSession: Error: invalid input: device 0x0, outputs 0xe7e7a850, callbacks 0xffbd8bb0, session 0xe7d12f9c
E/NdkCameraCaptureSession: ACameraCaptureSession_setRepeatingRequest: Error: invalid input: session 0x0, numRequest 1, requests 0xe7d12f90
D/ncnn: closeCamera
W/NdkCamera: close
D/HostConnection: HostConnection::get() New Host Connection established 0xe7b1bf30, tid 10018
D/HostConnection: HostComposition ext ANDROID_EMU_CHECKSUM_HELPER_v1 ANDROID_EMU_native_sync_v2 ANDROID_EMU_native_sync_v3 ANDROID_EMU_native_sync_v4 ANDROID_EMU_dma_v1 ANDROID_EMU_direct_mem ANDROID_EMU_vulkan ANDROID_EMU_deferred_vulkan_commands ANDROID_EMU_vulkan_null_optional_strings ANDROID_EMU_vulkan_create_resources_with_requirements ANDROID_EMU_YUV_Cache ANDROID_EMU_vulkan_ignored_handles ANDROID_EMU_has_shared_slots_host_memory_allocator ANDROID_EMU_vulkan_free_memory_sync ANDROID_EMU_vulkan_shader_float16_int8 ANDROID_EMU_vulkan_async_queue_submit ANDROID_EMU_sync_buffer_data ANDROID_EMU_vulkan_async_qsri ANDROID_EMU_read_color_buffer_dma GL_OES_EGL_image_external_essl3 GL_OES_vertex_array_object GL_KHR_texture_compression_astc_ldr ANDROID_EMU_host_side_tracing ANDROID_EMU_gles_max_version_3_0 
W/OpenGLRenderer: Failed to choose config with EGL_SWAP_BEHAVIOR_PRESERVED, retrying without...
D/EGL_emulation: eglCreateContext: 0xe7b1c6a0: maj 3 min 0 rcv 3
D/ncnn: setOutputWindow 0xe238c998
D/EGL_emulation: eglMakeCurrent: 0xe7b1c6a0: ver 3 0 (tinfo 0xe7e7b8b0) (first time)
I/Gralloc4: mapper 4.x is not supported
D/HostConnection: createUnique: call
D/HostConnection: HostConnection::get() New Host Connection established 0xe7b24890, tid 10018
D/goldfish-address-space: allocate: Ask for block of size 0x100
D/goldfish-address-space: allocate: ioctl allocate returned offset 0x3ebffe000 size 0x2000
D/HostConnection: HostComposition ext ANDROID_EMU_CHECKSUM_HELPER_v1 ANDROID_EMU_native_sync_v2 ANDROID_EMU_native_sync_v3 ANDROID_EMU_native_sync_v4 ANDROID_EMU_dma_v1 ANDROID_EMU_direct_mem ANDROID_EMU_vulkan ANDROID_EMU_deferred_vulkan_commands ANDROID_EMU_vulkan_null_optional_strings ANDROID_EMU_vulkan_create_resources_with_requirements ANDROID_EMU_YUV_Cache ANDROID_EMU_vulkan_ignored_handles ANDROID_EMU_has_shared_slots_host_memory_allocator ANDROID_EMU_vulkan_free_memory_sync ANDROID_EMU_vulkan_shader_float16_int8 ANDROID_EMU_vulkan_async_queue_submit ANDROID_EMU_sync_buffer_data ANDROID_EMU_vulkan_async_qsri ANDROID_EMU_read_color_buffer_dma GL_OES_EGL_image_external_essl3 GL_OES_vertex_array_object GL_KHR_texture_compression_astc_ldr ANDROID_EMU_host_side_tracing ANDROID_EMU_gles_max_version_3_0 
D/ncnn: openCamera 0
W/NdkCamera: open
W/NdkCamera: open 1 270
W/NdkCamera: onSessionActive 0xe7ddfb40
W/Gralloc4: allocator 3.x is not supported
W/ncnn: find_blob_index_by_name input failed
W/ncnn: Try
W/ncnn:     ex.input("images", in0);
A/libc: Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x0 in tid 10021 (ImageReader-640), pid 9990 (ncent.ncnnyolox)

[FeiGeChuanShu]

@hoanhvmetavi change the input to the real blob name in your param.
https://github.com/FeiGeChuanShu/ncnn-android-yolox/blob/main/app/src/main/jni/yolox.cpp#L357

[hoanhvmetavi]

@FeiGeChuanShu Forgive my dumb c++ skill. Thanks very much.

[hoanhvmetavi]

I was able to add a model to the app.
But when i add another new model that i trained, it does not have Yolov5 Focus layer. So i have to edit the model that following to this: https://github.com/Megvii-BaseDetection/YOLOX/tree/main/demo/ncnn/cpp
But in the step 4:

Then remove 10 lines of code from Split to Concat, but remember the last but 2nd number: 683.

I understand this step but my .param model file, position of the last but 2nd is not a number. Its /backbone/backbone/stem/Concat_output_0

Concat           /backbone/backbone/stem/Concat 4 1 /backbone/backbone/stem/Slice_1_output_0 /backbone/backbone/stem/Slice_4_output_0 /backbone/backbone/stem/Slice_2_output_0 /backbone/backbone/stem/Slice_5_output_0 /backbone/backbone/stem/Concat_output_0 0=0
Convolution      /backbone/backbone/stem/conv/conv/Conv 1 1 /backbone/backbone/stem/Concat_output_0 /backbone/backbone/stem/conv/conv/Conv_output_0 0=16 1=3 11=3 2=1 12=1 3=1 13=1 4=1 14=1 15=1 16=1 5=1 6=1728

So how do i convert this case, or am i missing another guide to handle this model?
Thanks.

[hoanhvmetavi]

For convenience, i will share my onnx model to you: https://1drv.ms/u/s!AoXQI1SRzY-tglL0qKI59iQcbibe?e=5mcq6I