Run or Build on Ubuntu 22.04 LTS fails

[drohhyn]

Hi, I am trying to get this running on Ubuntu 22.04 LTS. I've read the original google repo issues and also some info here, but I can't get it running. For me, whether installing wheels or compiling it myself fails or at least I am doing something wrong.

wheels

Installing wheels

python3 -m venv .
source bin/activate
python3 -m pip install pycoral-2.13.0-cp310-cp310-linux_x86_64.whl
python3 -m pip install tflite_runtime-2.13.0-cp310-cp310-linux_x86_64.whl

Installing tflite_runtime mentions tflite-runtime is already installed with the same version as the provided wheel. coming from the install of pycoral.

Running the example

Then I am trying to run the example from coral.ai (but from this repo tag 2.13.0), which is located in another directory (~/app/coral/pycoral), the wheels are located in ~/app/coralwhl.

python3 ~/app/coral/pycoral/examples/classify_image.py \
  --model ~/app/coral/pycoral/test_data/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite \
  --label ~/app/coral/pycoral/test_data/inat_bird_labels.txt \
  --input ~/app/coral/pycoral/test_data/parrot.jpg

It fails with this traceback:

Traceback (most recent call last):
  File "/home/USER/app/coral/pycoral/examples/classify_image.py", line 121, in <module>
    main()
  File "/home/USER/app/coral/pycoral/examples/classify_image.py", line 79, in main
    image = Image.open(args.input).convert('RGB').resize(size, Image.ANTIALIAS)
AttributeError: module 'PIL.Image' has no attribute 'ANTIALIAS'

Do I use the wrong libs?

Compiling

I've also tried to compile it myself after reading #85 and the issues in this repo. I've tried the master- and 2_14_0 branch, and also tag 2.13.0:

git clone https://github.com/oberluz/pycoral.git
cd pycoral
git checkout 1.13.0
git submodule init && git submodule update
sudo scripts/build.sh

(sudo because otherwise the connection to my docker instance is denied)
It fails after a while:

ERROR: /root/.cache/bazel/_bazel_root/eab0d61a99b6696edb3d2aff87b585e8/external/XNNPACK/BUILD.bazel:3204:19: Compiling src/amalgam/gen/neonfp16.c failed: (Exit 1): arm-linux-gnueabihf
-gcc failed: error executing command
  (cd /root/.cache/bazel/_bazel_root/eab0d61a99b6696edb3d2aff87b585e8/sandbox/processwrapper-sandbox/2759/execroot/pycoral && \
  exec env - \
    PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin \
    PWD=/proc/self/cwd \
  /usr/bin/arm-linux-gnueabihf-gcc -fPIC -U_FORTIFY_SOURCE -fstack-protector -Wall -Wunused-but-set-parameter -Wno-free-nonheap-object -fno-omit-frame-pointer '-march=armv7-a' '-mfpu=
neon-vfpv4' -g0 -O3 -DNDEBUG '-D_FORTIFY_SOURCE=2' -ffunction-sections -fdata-sections -funsafe-math-optimizations -ftree-vectorize '-std=c99' -MD -MF bazel-out/armv7a-opt/bin/externa
l/XNNPACK/_objs/neonfp16_amalgam_microkernels/neonfp16.d '-frandom-seed=bazel-out/armv7a-opt/bin/external/XNNPACK/_objs/neonfp16_amalgam_microkernels/neonfp16.o' '-DXNN_IGNORED_PLATFO
RM_JIT=0' -iquote external/XNNPACK -iquote bazel-out/armv7a-opt/bin/external/XNNPACK -isystem external/XNNPACK/include -isystem bazel-out/armv7a-opt/bin/external/XNNPACK/include -isys
tem external/XNNPACK/src -isystem bazel-out/armv7a-opt/bin/external/XNNPACK/src '-DNPY_NO_DEPRECATED_API=NPY_1_7_API_VERSION' '-ffp-contract=off' -Iinclude -Isrc -marm '-march=armv7-a
' '-mfpu=neon-fp16' '-std=c99' -O2 -no-canonical-prefixes -fno-canonical-system-headers -Wno-builtin-macro-redefined '-D__DATE__="redacted"' '-D__TIMESTAMP__="redacted"' '-D__TIME__="
redacted"' -c external/XNNPACK/src/amalgam/gen/neonfp16.c -o bazel-out/armv7a-opt/bin/external/XNNPACK/_objs/neonfp16_amalgam_microkernels/neonfp16.o)
# Configuration: 2e3e8616bf10e1d8f51b1c340571cfde3b0ecebcf0a5755d8726abf679f299e8
# Execution platform: @local_execution_config_platform//:platform

Use --sandbox_debug to see verbose messages from the sandbox and retain the sandbox build root for debugging
external/XNNPACK/src/amalgam/gen/neonfp16.c: In function 'xnn_f16_f32_vcvt_ukernel__neonfp16_x16':
external/XNNPACK/src/amalgam/gen/neonfp16.c:27:11: error: unknown type name 'float16x8_t'
     const float16x8_t vh0 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
           ^~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:27:29: warning: implicit declaration of function 'vreinterpretq_f16_u16'; did you mean 'vreinterpretq_s16_u16'? [-Wimplicit-function-declar
ation]
     const float16x8_t vh0 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
                             ^~~~~~~~~~~~~~~~~~~~~
                             vreinterpretq_s16_u16
external/XNNPACK/src/amalgam/gen/neonfp16.c:28:11: error: unknown type name 'float16x8_t'
     const float16x8_t vh1 = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
           ^~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:30:29: warning: implicit declaration of function 'vcvt_f32_f16'; did you mean 'vcvt_u32_f32'? [-Wimplicit-function-declaration]
     const float32x4_t vf0 = vcvt_f32_f16(vget_low_f16(vh0));
                             ^~~~~~~~~~~~
                             vcvt_u32_f32
external/XNNPACK/src/amalgam/gen/neonfp16.c:30:42: warning: implicit declaration of function 'vget_low_f16'; did you mean 'vget_low_p16'? [-Wimplicit-function-declaration]
     const float32x4_t vf0 = vcvt_f32_f16(vget_low_f16(vh0));
                                          ^~~~~~~~~~~~
                                          vget_low_p16
external/XNNPACK/src/amalgam/gen/neonfp16.c:30:29: error: incompatible types when initializing type 'float32x4_t' using type 'int'
     const float32x4_t vf0 = vcvt_f32_f16(vget_low_f16(vh0));
                             ^~~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:31:42: warning: implicit declaration of function 'vget_high_f16'; did you mean 'vget_high_p16'? [-Wimplicit-function-declaration]
     const float32x4_t vf1 = vcvt_f32_f16(vget_high_f16(vh0));
                                          ^~~~~~~~~~~~~
                                          vget_high_p16
external/XNNPACK/src/amalgam/gen/neonfp16.c:31:29: error: incompatible types when initializing type 'float32x4_t' using type 'int'
     const float32x4_t vf1 = vcvt_f32_f16(vget_high_f16(vh0));
                             ^~~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:32:29: error: incompatible types when initializing type 'float32x4_t' using type 'int'
     const float32x4_t vf2 = vcvt_f32_f16(vget_low_f16(vh1));
                             ^~~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:33:29: error: incompatible types when initializing type 'float32x4_t' using type 'int'
     const float32x4_t vf3 = vcvt_f32_f16(vget_high_f16(vh1));
                             ^~~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:41:11: error: unknown type name 'float16x8_t'
     const float16x8_t vh = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
           ^~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:43:31: error: incompatible types when initializing type 'float32x4_t' using type 'int'
     const float32x4_t vf_lo = vcvt_f32_f16(vget_low_f16(vh));
                               ^~~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:44:31: error: incompatible types when initializing type 'float32x4_t' using type 'int'
     const float32x4_t vf_hi = vcvt_f32_f16(vget_high_f16(vh));
                               ^~~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:52:11: error: unknown type name 'float16x8_t'
     const float16x8_t vh = vreinterpretq_f16_u16(vld1q_u16(i)); i += 8;
           ^~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:54:22: error: incompatible types when initializing type 'float32x4_t' using type 'int'
     float32x4_t vf = vcvt_f32_f16(vget_low_f16(vh));
                      ^~~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:57:10: error: incompatible types when assigning to type 'float32x4_t' from type 'int'
       vf = vcvt_f32_f16(vget_high_f16(vh));
          ^
external/XNNPACK/src/amalgam/gen/neonfp16.c: In function 'xnn_f32_f16_vcvt_ukernel__neonfp16_x16':
external/XNNPACK/src/amalgam/gen/neonfp16.c:88:28: warning: implicit declaration of function 'vreinterpretq_u16_f16'; did you mean 'vreinterpretq_u16_s16'? [-Wimplicit-function-declaration]
     const uint16x8_t vh0 = vreinterpretq_u16_f16(vcombine_f16(vcvt_f16_f32(vf0), vcvt_f16_f32(vf1)));
                            ^~~~~~~~~~~~~~~~~~~~~
                            vreinterpretq_u16_s16
external/XNNPACK/src/amalgam/gen/neonfp16.c:88:50: warning: implicit declaration of function 'vcombine_f16'; did you mean 'vcombine_p16'? [-Wimplicit-function-declaration]
     const uint16x8_t vh0 = vreinterpretq_u16_f16(vcombine_f16(vcvt_f16_f32(vf0), vcvt_f16_f32(vf1)));
                                                  ^~~~~~~~~~~~
                                                  vcombine_p16
external/XNNPACK/src/amalgam/gen/neonfp16.c:88:63: warning: implicit declaration of function 'vcvt_f16_f32'; did you mean 'vcvt_u32_f32'? [-Wimplicit-function-declaration]
     const uint16x8_t vh0 = vreinterpretq_u16_f16(vcombine_f16(vcvt_f16_f32(vf0), vcvt_f16_f32(vf1)));
                                                               ^~~~~~~~~~~~
                                                               vcvt_u32_f32
external/XNNPACK/src/amalgam/gen/neonfp16.c:88:28: error: incompatible types when initializing type 'uint16x8_t' using type 'int'
     const uint16x8_t vh0 = vreinterpretq_u16_f16(vcombine_f16(vcvt_f16_f32(vf0), vcvt_f16_f32(vf1)));
                            ^~~~~~~~~~~~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:89:28: error: incompatible types when initializing type 'uint16x8_t' using type 'int'
     const uint16x8_t vh1 = vreinterpretq_u16_f16(vcombine_f16(vcvt_f16_f32(vf2), vcvt_f16_f32(vf3)));
                            ^~~~~~~~~~~~~~~~~~~~~
external/XNNPACK/src/amalgam/gen/neonfp16.c:97:27: warning: implicit declaration of function 'vreinterpret_u16_f16'; did you mean 'vreinterpret_u16_s16'? [-Wimplicit-function-declaration]
     const uint16x4_t vh = vreinterpret_u16_f16(vcvt_f16_f32(vf));
                           ^~~~~~~~~~~~~~~~~~~~
                           vreinterpret_u16_s16
external/XNNPACK/src/amalgam/gen/neonfp16.c:97:27: error: incompatible types when initializing type 'uint16x4_t' using type 'int'
external/XNNPACK/src/amalgam/gen/neonfp16.c:107:21: error: incompatible types when initializing type 'uint16x4_t' using type 'int'
     uint16x4_t vh = vreinterpret_u16_f16(vcvt_f16_f32(vf));
                     ^~~~~~~~~~~~~~~~~~~~
Target //src:_pywrap_coral failed to build
INFO: Elapsed time: 21.970s, Critical Path: 17.28s
INFO: 474 processes: 182 internal, 292 processwrapper-sandbox.
FAILED: Build did NOT complete successfully
Makefile:154: recipe for target 'pybind' failed
make: *** [pybind] Error 1
make: Leaving directory '/workspace'
make: *** [/home/USER/app/coral/pycoral/libcoral/docker/docker.mk:40: docker-build] Error 1

Any ideas here?