快捷搜索： vl53l0 GPS户外定位机器人 物联网与机器人IORT GPS导航定位IORt机器人套件 log
首页 / 视频图像处理 /

教程4：TensorFlow2.0 label_image 的编译和使用

小v 4910 0

教程1：tensorflow lite 交叉编译和安装一 
教程2：tensorflow lite 编译和安装二 使用bazel编译  
教程3：tensorflow lite c++ 接口解读 
教程4：TensorFlow2.0 label_image 的编译和使用  

本教程是基于TensorFlow2.0的交叉编译。

使用的工具链是arm-linux-gnueabi-gcc 

32bit的。label_image是 一个目标识别开源程序。可以借用mobilenet的模型，对物体进行识别，识别可达上千种，识别的名单保存在一个叫mobilenet_v1.0_224_quant_labels.txt的文本里边，最终结合mobilenet_v1.0_224_quant.tflite 的量化或者 非量化模型mobilenet_v1.0_224.tflite进行使用.部分标签识别的名单如下： 


background
tench
goldfish
great white shark
tiger shark
hammerhead
electric ray
stingray
cock
hen
ostrich
brambling
goldfinch
house finch
junco
indigo bunting
robin
bulbul
jay
magpie
chickadee
water ouzel
kite
bald eagle
vulture
great grey owl
European fire salamander
common newt
eft
spotted salamander
axolotl
bullfrog
tree frog
tailed frog
loggerhead
leatherback turtle
mud turtle
terrapin
box turtle
banded gecko
common iguana
American chameleon
whiptail
agama
frilled lizard
alligator lizard
Gila monster
green lizard
African chameleon
Komodo dragon
African crocodile
American alligator
triceratops
thunder snake
ringneck snake
hognose snake
green snake
king snake
garter snake
water snake
vine snake
night snake
boa constrictor
rock python
Indian cobra
green mamba
sea snake
horned viper
diamondback
sidewinder
trilobite
harvestman
scorpion
black and gold garden spider
barn spider
garden spider
black widow
tarantula
wolf spider
tick
centipede
black grouse
ptarmigan
ruffed grouse
prairie chicken
peacock
quail
partridge
African grey
macaw
sulphur-crested cockatoo
lorikeet
coucal
bee eater
hornbill
hummingbird
jacamar
toucan
drake
red-breasted merganser
goose
black swan
tusker
echidna
platypus
wallaby
koala
wombat
jellyfish
sea anemone
brain coral
flatworm
nematode


默认编译TensorFlow lite  的时候，是不会编译label_image  的。所以需要修改。在路径tensorflow/lite/tools/make 

找到Makefile修改如下 


# Make uses /bin/sh by default, which is incompatible with the bashisms seen
# below.
SHELL := /bin/bash

# Find where we're running from, so we can store generated files here.
ifeq ($(origin MAKEFILE_DIR), undefined)
    MAKEFILE_DIR := $(shell dirname $(realpath $(lastword $(MAKEFILE_LIST))))
endif

# Try to figure out the host system

HOST_OS := linux

HOST_ARCH := armv7a

# Override these on the make command line to target a specific architecture. For example:
# make -f tensorflow/lite/tools/make/Makefile TARGET=rpi TARGET_ARCH=armv7l
TARGET := $(HOST_OS)
TARGET_ARCH := $(HOST_ARCH)

INCLUDES := \
-I. \
-I$(MAKEFILE_DIR)/../../../../../ \
-I$(MAKEFILE_DIR)/../../../../../../ \
-I$(MAKEFILE_DIR)/downloads/ \
-I$(MAKEFILE_DIR)/downloads/eigen \
-I$(MAKEFILE_DIR)/downloads/absl \
-I$(MAKEFILE_DIR)/downloads/gemmlowp \
-I$(MAKEFILE_DIR)/downloads/neon_2_sse \
-I$(MAKEFILE_DIR)/downloads/farmhash/src \
-I$(MAKEFILE_DIR)/downloads/flatbuffers/include \
-I$(OBJDIR)
# This is at the end so any globally-installed frameworks like protobuf don't
# override local versions in the source tree.
INCLUDES += -I/usr/local/include

# These are the default libraries needed, but they can be added to or
# overridden by the platform-specific settings in target makefiles.
LIBS := \
-lstdc++ \
-lpthread \
-lrt \
-lm \
-lz

# There are no rules for compiling objects for the host system (since we don't
# generate things like the protobuf compiler that require that), so all of
# these settings are for the target compiler.
CXXFLAGS := -O3 -DNDEBUG -fPIC
CXXFLAGS += $(EXTRA_CXXFLAGS)
CFLAGS := ${CXXFLAGS}
CXXFLAGS += --std=c++11
LDOPTS := 
ARFLAGS := -r
#TARGET_TOOLCHAIN_PREFIX :=
#CC_PREFIX :=

# This library is the main target for this makefile. It will contain a minimal
# runtime that can be linked in to other programs.
LIB_NAME := libtensorflow-lite.a

# Benchmark static library and binary
BENCHMARK_LIB_NAME := benchmark-lib.a
BENCHMARK_BINARY_NAME := benchmark_model

# A small example program that shows how to link against the library.
MINIMAL_SRCS := \
    tensorflow/lite/examples/minimal/minimal.cc
#add by lide
LABEL_IMAGE_SRCS := \
    tensorflow/lite/examples/label_image/label_image.cc \
    tensorflow/lite/examples/label_image/bitmap_helpers.cc 

# What sources we want to compile, must be kept in sync with the main Bazel
# build files.

PROFILER_SRCS := \
    tensorflow/lite/profiling/time.cc
PROFILE_SUMMARIZER_SRCS := \
    tensorflow/lite/profiling/profile_summarizer.cc \
    tensorflow/core/util/stats_calculator.cc

CMD_LINE_TOOLS_SRCS := \
    tensorflow/lite/tools/command_line_flags.cc

CORE_CC_ALL_SRCS := \
$(wildcard tensorflow/lite/*.cc) \
$(wildcard tensorflow/lite/*.c) \
$(wildcard tensorflow/lite/c/*.c) \
$(wildcard tensorflow/lite/core/*.cc) \
$(wildcard tensorflow/lite/core/api/*.cc) \
tensorflow/lite/experimental/resource_variable/*.cc \
tensorflow/lite/experimental/ruy/allocator.cc \
tensorflow/lite/experimental/ruy/block_map.cc \
tensorflow/lite/experimental/ruy/blocking_counter.cc \
tensorflow/lite/experimental/ruy/context.cc \
tensorflow/lite/experimental/ruy/detect_dotprod.cc \
tensorflow/lite/experimental/ruy/kernel_arm32.cc \
tensorflow/lite/experimental/ruy/kernel_arm64.cc \
tensorflow/lite/experimental/ruy/pack_arm.cc \
tensorflow/lite/experimental/ruy/pmu.cc \
tensorflow/lite/experimental/ruy/thread_pool.cc \
tensorflow/lite/experimental/ruy/trace.cc \
tensorflow/lite/experimental/ruy/trmul.cc \
tensorflow/lite/experimental/ruy/tune.cc \
tensorflow/lite/experimental/ruy/wait.cc
ifneq ($(BUILD_TYPE),micro)
CORE_CC_ALL_SRCS += \
$(wildcard tensorflow/lite/kernels/*.cc) \
$(wildcard tensorflow/lite/kernels/internal/*.cc) \
$(wildcard tensorflow/lite/kernels/internal/optimized/*.cc) \
$(wildcard tensorflow/lite/kernels/internal/reference/*.cc) \
$(PROFILER_SRCS) \
tensorflow/lite/tools/make/downloads/farmhash/src/farmhash.cc \
tensorflow/lite/tools/make/downloads/fft2d/fftsg.c \
tensorflow/lite/tools/make/downloads/flatbuffers/src/util.cpp
endif
# Remove any duplicates.add by lid
CORE_CC_ALL_SRCS := $(sort $(CORE_CC_ALL_SRCS))
CORE_CC_EXCLUDE_SRCS := \
$(wildcard tensorflow/lite/*test.cc) \
$(wildcard tensorflow/lite/*/*test.cc) \
$(wildcard tensorflow/lite/*/*/*test.cc) \
$(wildcard tensorflow/lite/*/*/*/*test.cc) \
$(wildcard tensorflow/lite/kernels/*test_main.cc) \
$(wildcard tensorflow/lite/kernels/*test_util.cc) \
$(MINIMAL_SRCS) \
$(LABEL_IMAGE_SRCS)
BUILD_WITH_MMAP ?= true
ifeq ($(BUILD_TYPE),micro)
    BUILD_WITH_MMAP=false
endif
ifeq ($(BUILD_TYPE),windows)
    BUILD_WITH_MMAP=false
endif
ifeq ($(BUILD_WITH_MMAP),true)
    CORE_CC_EXCLUDE_SRCS += tensorflow/lite/mmap_allocation.cc
else
    CORE_CC_EXCLUDE_SRCS += tensorflow/lite/mmap_allocation_disabled.cc
endif

BUILD_WITH_NNAPI ?= false
ifeq ($(BUILD_TYPE),micro)
    BUILD_WITH_NNAPI=false
endif
ifeq ($(TARGET),windows)
    BUILD_WITH_NNAPI=false
endif
ifeq ($(TARGET),ios)
    BUILD_WITH_NNAPI=false
endif
#nnapi false ->true add by lid
ifeq ($(TARGET),rpi)
    BUILD_WITH_NNAPI=false
endif
ifeq ($(TARGET),generic-aarch64)
    BUILD_WITH_NNAPI=false
endif
ifeq ($(BUILD_WITH_NNAPI),true)
    CORE_CC_ALL_SRCS += tensorflow/lite/delegates/nnapi/nnapi_delegate.cc
  CORE_CC_ALL_SRCS += tensorflow/lite/delegates/nnapi/quant_lstm_sup.cc
    CORE_CC_ALL_SRCS += tensorflow/lite/nnapi/nnapi_implementation.cc
    LIBS += -lrt
else
    CORE_CC_ALL_SRCS += tensorflow/lite/delegates/nnapi/nnapi_delegate_disabled.cc
    CORE_CC_ALL_SRCS += tensorflow/lite/nnapi/nnapi_implementation_disabled.cc
endif

ifeq ($(TARGET),ios)
    CORE_CC_EXCLUDE_SRCS += tensorflow/lite/minimal_logging_android.cc
    CORE_CC_EXCLUDE_SRCS += tensorflow/lite/minimal_logging_default.cc
else
    CORE_CC_EXCLUDE_SRCS += tensorflow/lite/minimal_logging_android.cc
    CORE_CC_EXCLUDE_SRCS += tensorflow/lite/minimal_logging_ios.cc
endif

# Filter out all the excluded files.
TF_LITE_CC_SRCS := $(filter-out $(CORE_CC_EXCLUDE_SRCS), $(CORE_CC_ALL_SRCS))

# Benchmark sources
BENCHMARK_SRCS_DIR := tensorflow/lite/tools/benchmark
EVALUATION_UTILS_SRCS := \
  tensorflow/lite/tools/evaluation/utils.cc
BENCHMARK_ALL_SRCS := $(TF_LITE_CC_SRCS) \
    $(wildcard $(BENCHMARK_SRCS_DIR)/*.cc) \
    $(PROFILE_SUMMARIZER_SRCS) \
    $(CMD_LINE_TOOLS_SRCS) \
    $(EVALUATION_UTILS_SRCS)

BENCHMARK_SRCS := $(filter-out \
    $(wildcard $(BENCHMARK_SRCS_DIR)/*_test.cc) \
    $(BENCHMARK_SRCS_DIR)/benchmark_plus_flex_main.cc, \
    $(BENCHMARK_ALL_SRCS))

# These target-specific makefiles should modify or replace options like
# CXXFLAGS or LIBS to work for a specific targetted architecture. All logic
# based on platforms or architectures should happen within these files, to
# keep this main makefile focused on the sources and dependencies.
include $(wildcard $(MAKEFILE_DIR)/targets/*_makefile.inc)
#add by lid
ALL_SRCS := \
    $(MINIMAL_SRCS) \
        $(LABEL_IMAGE_SRCS) \
    $(PROFILER_SRCS) \
    $(PROFILER_SUMMARIZER_SRCS) \
    $(TF_LITE_CC_SRCS) \
    $(BENCHMARK_SRCS) \
    $(CMD_LINE_TOOLS_SRCS)

# Where compiled objects are stored.
GENDIR := $(MAKEFILE_DIR)/gen/$(TARGET)_$(TARGET_ARCH)/
OBJDIR := $(GENDIR)obj/
BINDIR := $(GENDIR)bin/
LIBDIR := $(GENDIR)lib/
#ADD by lid
LIB_PATH := $(LIBDIR)$(LIB_NAME)
BENCHMARK_LIB := $(LIBDIR)$(BENCHMARK_LIB_NAME)
BENCHMARK_BINARY := $(BINDIR)$(BENCHMARK_BINARY_NAME)
MINIMAL_BINARY := $(BINDIR)minimal
LABEL_IMAGE_BINARY :=$(BINDIR)label_image

CXX := $(CC_PREFIX)${TARGET_TOOLCHAIN_PREFIX}g++
CC := $(CC_PREFIX)${TARGET_TOOLCHAIN_PREFIX}gcc
AR := $(CC_PREFIX)${TARGET_TOOLCHAIN_PREFIX}ar

MINIMAL_OBJS := $(addprefix $(OBJDIR), \
$(patsubst %.cc,%.o,$(patsubst %.c,%.o,$(MINIMAL_SRCS))))

LIB_OBJS := $(addprefix $(OBJDIR), \
$(patsubst %.cc,%.o,$(patsubst %.c,%.o,$(patsubst %.cpp,%.o,$(TF_LITE_CC_SRCS)))))

BENCHMARK_OBJS := $(addprefix $(OBJDIR), \
$(patsubst %.cc,%.o,$(patsubst %.c,%.o,$(patsubst %.cpp,%.o,$(BENCHMARK_SRCS)))))
#ADD by lid
LABEL_IMAGE_OBJS := $(addprefix $(OBJDIR), \
$(patsubst %.cc,%.o,$(patsubst %.c,%.o,$(LABEL_IMAGE_SRCS))))

# For normal manually-created TensorFlow Lite C++ source files.
$(OBJDIR)%.o: %.cc
    @mkdir -p $(dir $@)
    $(CXX) $(CXXFLAGS) $(INCLUDES) -c $< -o $@
# For normal manually-created TensorFlow Lite C source files.
$(OBJDIR)%.o: %.c
    @mkdir -p $(dir $@)
    $(CC) $(CFLAGS) $(INCLUDES) -c $< -o $@
$(OBJDIR)%.o: %.cpp
    @mkdir -p $(dir $@)
    $(CXX) $(CXXFLAGS) $(INCLUDES) -c $< -o $@

# The target that's compiled if there's no command-line arguments.
all: $(LIB_PATH)  $(MINIMAL_BINARY) $(BENCHMARK_BINARY) $(LABEL_IMAGE_BINARY)

# The target that's compiled for micro-controllers
micro: $(LIB_PATH)

# Hack for generating schema file bypassing flatbuffer parsing
tensorflow/lite/schema/schema_generated.h:
    @cp -u tensorflow/lite/schema/schema_generated.h.OPENSOURCE tensorflow/lite/schema/schema_generated.h

# Gathers together all the objects we've compiled into a single '.a' archive.
$(LIB_PATH): tensorflow/lite/schema/schema_generated.h $(LIB_OBJS)
    @mkdir -p $(dir $@)
    $(AR) $(ARFLAGS) $(LIB_PATH) $(LIB_OBJS)

lib: $(LIB_PATH)

$(MINIMAL_BINARY): $(MINIMAL_OBJS) $(LIB_PATH)
    @mkdir -p $(dir $@)
    $(CXX) $(CXXFLAGS) $(INCLUDES) \
    -o $(MINIMAL_BINARY) $(MINIMAL_OBJS) \
    $(LIBFLAGS) $(LIB_PATH) $(LDFLAGS) $(LIBS)

minimal: $(MINIMAL_BINARY)

#ADD by lide
$(LABEL_IMAGE_BINARY): $(LABEL_IMAGE_OBJS) $(LIB_PATH)
    @mkdir -p $(dir $@)
    $(CXX) $(CXXFLAGS) $(INCLUDES) \
    -o $(LABEL_IMAGE_BINARY) $(LABEL_IMAGE_OBJS) \
    $(LIBFLAGS) $(LIB_PATH) $(LDFLAGS) $(LIBS)
label_image:$(LABEL_IMAGE_BINARY)

$(BENCHMARK_LIB) : $(LIB_PATH) $(BENCHMARK_OBJS)
    @mkdir -p $(dir $@)
    $(AR) $(ARFLAGS) $(BENCHMARK_LIB) $(LIB_OBJS) $(BENCHMARK_OBJS)

benchmark_lib: $(BENCHMARK_LIB)

$(BENCHMARK_BINARY) : $(BENCHMARK_LIB)
    @mkdir -p $(dir $@)
    $(CXX) $(CXXFLAGS) $(INCLUDES) \
    -o $(BENCHMARK_BINARY) \
    $(LIBFLAGS) $(BENCHMARK_LIB) $(LDFLAGS) $(LIBS)

benchmark: $(BENCHMARK_BINARY)

libdir:
    @echo $(LIBDIR)

# Gets rid of all generated files.
clean:
    rm -rf $(MAKEFILE_DIR)/gen

# Gets rid of target files only, leaving the host alone. Also leaves the lib
# directory untouched deliberately, so we can persist multiple architectures
# across builds for iOS and Android.
cleantarget:
    rm -rf $(OBJDIR)
    rm -rf $(BINDIR)

$(DEPDIR)/%.d: ;
.PRECIOUS: $(DEPDIR)/%.d

-include $(patsubst %,$(DEPDIR)/%.d,$(basename $(ALL_SRCS)))


added  by lid 是我修改后的内容。然而在编译 ./build_rpi.sh 的时候，仍旧会报错。 


examples/label_image/label_image.o /home/lid/tensorflow_lite/tensorflow/tensorflow/lite/tools/make/gen/rpi_armv7l/obj/tensorflow/lite/examples/label_image/bitmap_helpers.o \
     /home/lid/tensorflow_lite/tensorflow/tensorflow/lite/tools/make/gen/rpi_armv7l/lib/libtensorflow-lite.a -Wl,--no-export-dynamic -Wl,--exclude-libs,ALL -Wl,--gc-sections -Wl,--as-needed -lstdc++ -lpthread -lm -ldl -lrt
/home/lid/tensorflow_lite/tensorflow/tensorflow/lite/tools/make/gen/rpi_armv7l/obj/tensorflow/lite/examples/label_image/label_image.o: In function `tflite::label_image::CreateGPUDelegate(tflite::label_image::Settings*)':
label_image.cc:(.text+0x2a): undefined reference to `tflite::evaluation::CreateGPUDelegate(tflite::FlatBufferModel*)'
/home/lid/tensorflow_lite/tensorflow/tensorflow/lite/tools/make/gen/rpi_armv7l/obj/tensorflow/lite/examples/label_image/label_image.o: In function `tflite::label_image::GetDelegates(tflite::label_image::Settings*)':
label_image.cc:(.text+0x632): undefined reference to `tflite::evaluation::CreateNNAPIDelegate()'
label_image.cc:(.text+0x63e): undefined reference to `tflite::evaluation::CreateNNAPIDelegate()'
collect2: error: ld returned 1 exit status
make: *** [/home/lid/tensorflow_lite/tensorflow/tensorflow/lite/tools/make/gen/rpi_armv7l/bin/label_image] Error 1
make: Leaving directory `/home/lid/tensorflow_lite/tensorflow'


 找到label的路径




/tensorflow/lite/examples/label_image

label_image.cc






将73行如下修改


//return evaluation::CreateGPUDelegate(s->model);


87行如下修改




#if 0
  if (s->accel) {
    auto delegate = evaluation::CreateNNAPIDelegate();
    if (!delegate) {
      LOG(INFO) << "NNAPI acceleration is unsupported on this platform.";
    } else {
      delegates.emplace("NNAPI", evaluation::CreateNNAPIDelegate());
    }
  }
#endif 

保存，回到 make 的路径下，./build_rpi.sh  ，编译成功不再报错。将 label_image、grace

e_hopper.bmp  、

mobilenet_v1_1.0_224_quant.tflite 、

labels_mobilenet_quant_v1_224.txt、






传到arm板子上。




执行 


./label_image -v 4 -m ./mobilenet_v1_1.0_224_quant.tflite -i ./grac
e_hopper.bmp    -l labels_mobilenet_quant_v1_224.txt
输出结果如下：



88个input 87个output


average time: 420.349 ms 
0.780392: 653name: military uniform
0.105882: 907name: Windsor tie
0.0156863: 458name: bow tie
0.0117647: 466name: bulletproof vest
0.00784314: 835name: suit









    换个试试




    微信截图_20191226112256.png 




    





    average time: 413.936 ms 

0.207843: 950name: strawberry

0.105882: 748name: punching bag

0.0784314: 761name: refrigerator

0.0666667: 418name: balloon

0.0470588: 729name: plastic bag




    





    


				

				

					
					视频图像处理				

								

							
		
		
					

				

					
 您可能还会对下面的文章感兴趣：

					
    
											

				

				

					
 分类文章

					
				

				

			
	

		
		

			
		

				

			

				
			

			

				
			

			

				
			

			

				 
			

		

				

			
		

		
	



			

		

	

	



	





sitemap