Do NOT require Flatbuffers (#4614)
author박종현/On-Device Lab(SR)/Staff Engineer/삼성전자 <jh1302.park@samsung.com>
Wed, 6 Mar 2019 11:36:58 +0000 (20:36 +0900)
committer오형석/On-Device Lab(SR)/Staff Engineer/삼성전자 <hseok82.oh@samsung.com>
Wed, 6 Mar 2019 11:36:58 +0000 (20:36 +0900)
It turns out that TensorFlow Lite does not depend on Flatbuffers
runtime.

Signed-off-by: Jonghyun Park <jh1302.park@samsung.com>
externals/CMakeLists.txt

index 2644984..b51aa45 100644 (file)
@@ -51,8 +51,6 @@ list(APPEND TFLITE_INCLUDES "${TFLITE_DEPEND_DIR}/farmhash/src")
 list(APPEND TFLITE_INCLUDES "${TFLITE_DEPEND_DIR}/flatbuffers/include")
 
 if(BUILD_IS_NATIVE AND NOT HOST_ARCH_BASE STREQUAL "arm")
-  # Flatbuffer build
-  nnfw_find_package(FlatBuffers)
   # Required external sourcefor x86-64 build
   nnfw_find_package(NEON2SSESource REQUIRED)
   list(APPEND TFLITE_INCLUDES "${TFLITE_DEPEND_DIR}/neon_2_sse")