为 Android 构建 TensorFlow Lite 时出现问题

Question

我在尝试为 Android 构建 TensorFlow Lite 时遇到很多错误。我正在使用代码遵循https://www.tensorflow.org/lite/guide/ops_select 的指导方针：

bazel build --cxxopt='--std=c++11' -c opt             \
  --config=android_arm --config=monolithic          \
  //tensorflow/lite/java:tensorflow-lite-with-select-tf-ops

甚至尝试过：

bazel build --cxxopt='--std=c++11' -c opt             \
  --config=android_arm64 --config=monolithic          \
  //tensorflow/lite/java:tensorflow-lite-with-select-tf-ops

但我收到如下错误：

Execution platform: @bazel_tools//platforms:host_platform
external/com_google_absl/absl/synchronization/mutex.cc:1103:1: error: 'xray_log_args' attribute is invalid for the implicit this argument
ABSL_XRAY_LOG_ARGS(1) void Mutex::Block(PerThreadSynch *s) {
^                  ~
external/com_google_absl/absl/base/attributes.h:514:38: note: expanded from macro 'ABSL_XRAY_LOG_ARGS'
    [[clang::xray_always_instrument, clang::xray_log_args(N)]]
                                     ^                    ~
external/com_google_absl/absl/synchronization/mutex.cc:1450:1: error: 'xray_log_args' attribute is invalid for the implicit this argument
ABSL_XRAY_LOG_ARGS(1) void Mutex::Lock() {
^                  ~
external/com_google_absl/absl/base/attributes.h:514:38: note: expanded from macro 'ABSL_XRAY_LOG_ARGS'
    [[clang::xray_always_instrument, clang::xray_log_args(N)]]
                                     ^                    ~
external/com_google_absl/absl/synchronization/mutex.cc:1468:1: error: 'xray_log_args' attribute is invalid for the implicit this argument
ABSL_XRAY_LOG_ARGS(1) void Mutex::ReaderLock() {
^                  ~
external/com_google_absl/absl/base/attributes.h:514:38: note: expanded from macro 'ABSL_XRAY_LOG_ARGS'
    [[clang::xray_always_instrument, clang::xray_log_args(N)]]
                                     ^                    ~
external/com_google_absl/absl/synchronization/mutex.cc:1581:1: error: 'xray_log_args' attribute is invalid for the implicit this argument
ABSL_XRAY_LOG_ARGS(1) bool Mutex::TryLock() {
^                  ~
external/com_google_absl/absl/base/attributes.h:514:38: note: expanded from macro 'ABSL_XRAY_LOG_ARGS'
    [[clang::xray_always_instrument, clang::xray_log_args(N)]]
                                     ^                    ~
external/com_google_absl/absl/synchronization/mutex.cc:1610:1: error: 'xray_log_args' attribute is invalid for the implicit this argument
ABSL_XRAY_LOG_ARGS(1) bool Mutex::ReaderTryLock() {
^                  ~
external/com_google_absl/absl/base/attributes.h:514:38: note: expanded from macro 'ABSL_XRAY_LOG_ARGS'
    [[clang::xray_always_instrument, clang::xray_log_args(N)]]
                                     ^                    ~
external/com_google_absl/absl/synchronization/mutex.cc:1656:1: error: 'xray_log_args' attribute is invalid for the implicit this argument
ABSL_XRAY_LOG_ARGS(1) void Mutex::Unlock() {
^                  ~
external/com_google_absl/absl/base/attributes.h:514:38: note: expanded from macro 'ABSL_XRAY_LOG_ARGS'
    [[clang::xray_always_instrument, clang::xray_log_args(N)]]
                                     ^                    ~
external/com_google_absl/absl/synchronization/mutex.cc:1708:1: error: 'xray_log_args' attribute is invalid for the implicit this argument
ABSL_XRAY_LOG_ARGS(1) void Mutex::ReaderUnlock() {
^                  ~
external/com_google_absl/absl/base/attributes.h:514:38: note: expanded from macro 'ABSL_XRAY_LOG_ARGS'
    [[clang::xray_always_instrument, clang::xray_log_args(N)]]
                                     ^                    ~
7 errors generated.
Target //tensorflow/lite/java:tensorflow-lite-with-select-tf-ops failed to build
INFO: Elapsed time: 555.749s, Critical Path: 174.46s
INFO: 490 processes: 485 local, 5 worker.
FAILED: Build did NOT complete successfully

还尝试关注https://becominghuman.ai/how-to-build-tensorflow-as-a-static-library-for-android-5c762dbdd5d4 创建静态库，但随后在其他文件中出现不同类型的错误

 error: 'to_string' is not a member of 'std'

我的构建系统是 Ubuntu，我在 r1.14 分支中

有没有人设法构建适用于 Android 的 TensorFlow Lite，可以给我一些建议？

Answer 1

我不知道第一组错误，但我得到了第二个错误，然后尝试使用旧的 NDK 版本编译 TFLite。尝试使用 Android NDK Revision 18b，不要忘记重新运行 configure 脚本。