Question

我希望在RPi3上安装TF Lite，以进行实时图像识别。

我关注了：https://github.com/tensorflow/tensorflow/blob/master/tensorflow/contrib/lite/g3doc/rpi.md

所有已编译的libtensorflow-lite.a成功构建。

尝试运行初始示例应用程序：

/etc/hosts

我得到：

python label_image.py

我不确定的是：

我错过了什么。
确保TF使用lite库而不是完整版时需要做的事情。
最后-没人知道Pi上应该实现什么FPS TF lite吗？

Answer 1

对不起，我不了解Stack Overflow的规则。
我修改了评论，但如果评论者不喜欢，请删除它。

可以通过官方pip命令引入的“ Tensorflow v1.11.0”引起了问题。

undefined symbol: _ZN6tflite12tensor_utils39NeonMatrixBatchVectorMultiplyAccumulateEPKaiiS2_PKfiPfi

点子包装似乎已损坏。

Tensorflow Lite的构建过程如下。

1。执行下面的

$ sudo apt update;sudo apt upgrade -y
$ sudo apt-get install -y build-essential openjdk-8-jdk pkg-config 
zip g++ zlib1g-dev unzip

2。重启终端。

3。执行以下操作。（需要运行两次）

$ sudo apt-get install -y openjdk-8-jdk

4.Bazel构建。

$ cd ~
$ mkdir bazel;cd bazel
$ wget https://github.com/bazelbuild/bazel/releases/download/0.17.2/bazel-0.17.2-dist.zip
$ unzip bazel-0.17.2-dist.zip
$ nano bazel/scripts/bootstrap/compile.sh

run "${JAVAC}" -classpath "${classpath}" -sourcepath "${sourcepath}" \
-d "${output}/classes" -source "$JAVA_VERSION" -target "$JAVA_VERSION" \
-encoding UTF-8 ${BAZEL_JAVAC_OPTS} "@${paramfile}"
↓
run "${JAVAC}" -classpath "${classpath}" -sourcepath "${sourcepath}" \
-d "${output}/classes" -source "$JAVA_VERSION" -target "$JAVA_VERSION" \
-encoding UTF-8 ${BAZEL_JAVAC_OPTS} "@${paramfile}" -J-Xmx500M

$ sudo bash ./compile.sh
$ sudo cp output/bazel /usr/local/bin

5。为Tensorflow构建做准备。

$ cd ~
$ sudo pip2 uninstall tensorflow
$ sudo pip3 uninstall tensorflow
$ sudo pip3 install keras_applications==1.0.4 --no-deps
$ sudo pip3 install keras_preprocessing==1.0.2 --no-deps
$ sudo pip3 install h5py==2.8.0
$ git clone -b v1.11.0 https://github.com/tensorflow/tensorflow.git
$ cd tensorflow
$ git checkout v1.11.0
$ ./tensorflow/contrib/lite/tools/make/download_dependencies.sh
$ ./tensorflow/contrib/lite/tools/make/build_rpi_lib.sh
$ sudo bazel build tensorflow/contrib/lite/toco:toco

6-1.Tensorflow的配置。（Python2.x）

$ ./configure
WARNING: Running Bazel server needs to be killed, because the startup options are different.
WARNING: --batch mode is deprecated. Please instead explicitly shut down your Bazel server using the command "bazel shutdown".
You have bazel 0.17.2- (@non-git) installed.
Please specify the location of python. [Default is /usr/bin/python]: 
Found possible Python library paths:
/usr/local/lib/python2.7/dist-packages
/usr/local/lib
/usr/lib/python2.7/dist-packages
/opt/movidius/caffe/python
Please input the desired Python library path to use.  Default is [/usr/local/lib/python2.7/dist-packages]

Do you wish to build TensorFlow with jemalloc as malloc support? [Y/n]: n
No jemalloc as malloc support will be enabled for TensorFlow.

Do you wish to build TensorFlow with Google Cloud Platform support? [Y/n]: n
No Google Cloud Platform support will be enabled for TensorFlow.

Do you wish to build TensorFlow with Hadoop File System support? [Y/n]: n
No Hadoop File System support will be enabled for TensorFlow.

Do you wish to build TensorFlow with Amazon AWS Platform support? [Y/n]: n
No Amazon AWS Platform support will be enabled for TensorFlow.

Do you wish to build TensorFlow with Apache Kafka Platform support? [Y/n]: n
No Apache Kafka Platform support will be enabled for TensorFlow.

Do you wish to build TensorFlow with XLA JIT support? [y/N]: n
No XLA JIT support will be enabled for TensorFlow.

Do you wish to build TensorFlow with GDR support? [y/N]: n
No GDR support will be enabled for TensorFlow.

Do you wish to build TensorFlow with VERBS support? [y/N]: n
No VERBS support will be enabled for TensorFlow.

Do you wish to build TensorFlow with nGraph support? [y/N]: n
No nGraph support will be enabled for TensorFlow.

Do you wish to build TensorFlow with OpenCL SYCL support? [y/N]: n
No OpenCL SYCL support will be enabled for TensorFlow.

Do you wish to build TensorFlow with CUDA support? [y/N]: n
No CUDA support will be enabled for TensorFlow.

Do you wish to download a fresh release of clang? (Experimental) [y/N]: n
Clang will not be downloaded.

Do you wish to build TensorFlow with MPI support? [y/N]: n
No MPI support will be enabled for TensorFlow.

Please specify optimization flags to use during compilation when bazel option "--config=opt" is specified [Default is -march=native]: 


Would you like to interactively configure ./WORKSPACE for Android builds? [y/N]: n
Not configuring the WORKSPACE for Android builds.

Preconfigured Bazel build configs. You can use any of the below by adding "--config=<>" to your build command. See tools/bazel.rc for more details.
--config=mkl            # Build with MKL support.
--config=monolithic     # Config for mostly static monolithic build.
Configuration finished

6-2.Tensorflow的配置。（Python3.x）

Please specify the location of python. [Default is /usr/bin/python]: /usr/bin/python3


Found possible Python library paths:
  /usr/local/lib
  /usr/lib/python3/dist-packages
  /usr/local/lib/python3.5/dist-packages
Please input the desired Python library path to use.  Default is [/usr/local/lib] /usr/local/lib/python3.5/dist-packages

7.Tensorflow的构建。

$ sudo bazel build --config opt --local_resources 1024.0,0.5,0.5 \
--copt=-mfpu=neon-vfpv4 \
--copt=-ftree-vectorize \
--copt=-funsafe-math-optimizations \
--copt=-ftree-loop-vectorize \
--copt=-fomit-frame-pointer \
--copt=-DRASPBERRY_PI \
--host_copt=-DRASPBERRY_PI \
//tensorflow/tools/pip_package:build_pip_package

8.Tensorflow的安装。

$ sudo ./bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg
$ sudo pip2 install /tmp/tensorflow_pkg/tensorflow-1.11.0-cp27-cp27mu-linux_armv7l.whl
Or
$ sudo pip3 install /tmp/tensorflow_pkg/tensorflow-1.11.0-cp35-cp35m-linux_armv7l.whl

9。【必需】重启终端。

10。执行下面的

$ sudo apt install -y python-scipy python3-scipy
$ sudo apt remove openjdk-8-*
$ sudo apt purge openjdk-8-*

如果到目前为止已完成，则该问题应该已经解决。
但是，总共需要27个小时。
如果您不等待27个小时，则可以使用交叉构建或使用建议的滚轮文件大大减少等待时间。

预构建Tensorflow二进制文件（车轮）在下面。
https://github.com/PINTO0309/Tensorflow-bin

Answer 2

我错过了什么。

要在Raspberry Pi上安装Tensorflow，您可以使用官方的pip包

pip install tensorflow

有关更多详细信息，请访问官方文档页面-

https://www.tensorflow.org/install/pip

或者，如果需要Cross编译的软件包，则可以遵循此官方文档页面-

https://www.tensorflow.org/install/source_rpi

为确保TF使用lite库而不是完整版，我需要做些什么。

您只需要一个经过训练的模型并将其转换为TFLite模型即可。有关完整的详细信息，您可以按照本教程进行操作-

https://medium.com/tensorflow/training-and-serving-a-realtime-mobile-object-detector-in-30-minutes-with-cloud-tpus-b78971cf1193

最后-没人知道Pi上应该实现什么FPS TF lite吗？

我目前正在研究它，但是由于tensorflow当前针对Raspberry pi（https://github.com/tensorflow/tensorflow/issues/23082）的官方构建存在一些问题，因此我无法计算Pi的性能。

Answer 3

要关注Ishita Shah的评论，我在这里发布我的步骤，以构建/运行在Raspberry Pi 3上运行的TfLite的2个启动应用程序：-）

1，按照https://www.raspberrypi.org/learning/software-guide/quickstart/，将Raspbian安装到Raspberry Pi 3板上。

#sudo apt install build-essential 按照https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/pi_examples安装摄像头。验证相机是否正常工作

#raspistill -v

2，git clone -b r1.12 https://github.com/huaxiaozhong1/tensorflow.git。（最好的办法是在Ubuntu 18.04的最新docker容器中完成）。它将6个静态库克隆到/ tensorflow / tensorflow / contrib / lite / examples / camera / libs中： libjpeg.a librt.a libv4l1.a libv4l2.a libv4l2rds.a libv4lconvert.a。实际上，它们是从在步骤1中准备的Pi环境中创建的：

3，在主机容器中：

#apt update

#apt upgrade

#apt install crossbuild-essential-armhf

#cd /tensorflow

#./tensorflow/contrib/lite/tools/make/download_dependencies.sh

#apt install -y libjpeg-dev

#apt install libv4l-dev

#./tensorflow/contrib/lite/tools/make/build_rpi_lib.sh

现在，您可以找到./tensorflow/contrib/lite/tools/make/gen/rpi_armv7l/lib/libtensorflow-lite.a、/tensorflow/tensorflow/contrib/lite/tools/make/gen/rpi_armv7l/bin / label_image和/ tensorflow / tensorflow / contrib / lite / tools / make / gen / rpi_armv7l / bin / camera。

4，将label_image和camera（Linux可执行文件）复制到Raspeberry板上。从主机容器的/ tensorflow / tensorflow / contrib / lite / examples / label_image / testdata复制grace_hopper.bmp到板上。从主机容器的/ tensorflow / tensorflow / contrib / lite / java / ovic / src / testdata复制label.txt到开发板上。从https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/g3doc/models.md将Mobilenet_v2_1.0_224_quant.tflite下载到开发板。 3个数据文件和2个Linux可执行文件应位于同一文件夹中。

5，一个接一个地运行两个应用程序。

注意：我将2个应用程序添加到的基本分支是r1.12。现在，TfLite已宣布退出竞争。但是，当我在最新的tensorflow / tensorflow：nightly-devel中运行build_rpi_lib.sh时，它没有成功。那么，我现在可以基于r1.12提交拉取请求吗？成功构建libtensorflow-lite.a后，我可以在r1.13 +上添加相同的功能。

Tensor Flow Lite Raspberry Pi安装

3 个答案: