terryky / tflite_gles_app

GPU accelerated deep learning inference applications for RaspberryPi / JetsonNano / Linux PC using TensorflowLite GPUDelegate / TensorRT
MIT License
495 stars 131 forks source link

Cannot reproduce handpose in x86 platform #1

Closed AlexTraan closed 4 years ago

AlexTraan commented 4 years ago

Hi Author,

Thank you for your valuable repository.

I am trying to reproduce your handpose for x86 platform, but when i build, i faced error as below: "In file included from tflite_handpose.cpp:5:0: ../common/util_tflite.h:8:46: fatal error: tensorflow/lite/kernels/register.h: No such file or directory compilation terminated. ../Makefile.include:18: recipe for target 'tflite_handpose.o' failed make: [tflite_handpose.o] Error 1 make: Waiting for unfinished jobs.... In file included from custom_ops/transpose_conv_bias.cc:19:0: custom_ops/transpose_conv_bias.h:18:52: fatal error: tensorflow/lite/kernels/internal/types.h: No such file or directory compilation terminated. ../Makefile.include:21: recipe for target 'custom_ops/transpose_conv_bias.o' failed make: *** [custom_ops/transpose_conv_bias.o] Error 1 In file included from ../common/util_tflite.cpp:5:0: ../common/util_tflite.h:8:46: fatal error: tensorflow/lite/kernels/register.h: No such file or directory"

It quite strange when that path ( tensorflow/lite/kernels/register.h,...) existed.

Can you help?

terryky commented 4 years ago

Hi AlexTraan san,

Did you run the script below ?

$ cd ~/work
$ ./tflite_gles_app/tools/scripts/tf2.2/build_libtflite_r2.2.sh
$ ln -s tensorflow_r2.2 ./tensorflow

And then, please confirm if the header file exists at ~/work/tensorflow/tensorflow/lite/kernels/register.h.

AlexTraan commented 4 years ago

Hi Yoshiteru san,

Thank you for your feedback.

"Did you run the script below ?"

yes

"And then, please confirm if the header file exists at ~/work/tensorflow/tensorflow/lite/kernels/register.h."

I forget to creat "~/work" folder. I will try again.

Sorry for my uncareful reading.

terryky commented 4 years ago

Please reopen if problem still exists.