Update for README.md files for ios_examples and makefile.
Change: 125601282
This commit is contained in:
A. Unique TensorFlower
TensorFlower Gardener
parent
7614900108
commit
e231a8b382
@ -6,38 +6,71 @@ This folder contains examples of how to build applications for iOS devices using
|
||||
|
||||
- You'll need Xcode 7.3 or later, with the command-line tools installed.
|
||||
|
||||
- Follow the instructions at [tensorflow/contrib/makefile](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/makefile) to compile a static library containing the core TensorFlow code.
|
||||
- Follow the instructions at
|
||||
[tensorflow/contrib/makefile](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/makefile)
|
||||
under "iOS" to compile a static library containing the core TensorFlow code.
|
||||
|
||||
- Download [Inception v1](https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip), and extract the label and graph files into the data folders inside both the simple and camera examples.
|
||||
- Download
|
||||
[Inception v1](https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip),
|
||||
and extract the label and graph files into the data folders inside both the
|
||||
simple and camera examples.
|
||||
|
||||
- Load the Xcode project inside the `simple` subfolder, and press Command-R to build and run it on the simulator or your connected device.
|
||||
- Load the Xcode project inside the `simple` subfolder, and press Command-R to
|
||||
build and run it on the simulator or your connected device.
|
||||
|
||||
- You should see a single-screen app with a "Run Model" button. Tap that, and you should see some debug output appear below indicating that the example Grace Hopper image has been analyzed, with a military uniform recognized.
|
||||
- You should see a single-screen app with a "Run Model" button. Tap that, and
|
||||
you should see some debug output appear below indicating that the example
|
||||
Grace Hopper image has been analyzed, with a military uniform recognized.
|
||||
|
||||
- Once that's been successfully run, make sure you have a real device connected and open up the Xcode project in the camera subfolder. Once you build and run that, you should get a live camera view that you can point at objects to get real-time recognition results.
|
||||
- Once you have success there, make sure you have a real device connected and
|
||||
open up the Xcode project in the camera subfolder. Once you build and run
|
||||
that, you should get a live camera view that you can point at objects to get
|
||||
real-time recognition results.
|
||||
|
||||
## Troubleshooting
|
||||
|
||||
If you're hitting problems, here's a checklist of common things to investigate:
|
||||
|
||||
- Make sure that you've run the `download_dependencies.sh` and `compile_ios_protobuf.sh` scripts before you run `compile_ios_tensorflow`.
|
||||
- Make sure that you've run the `download_dependencies.sh` and
|
||||
`compile_ios_protobuf.sh` scripts before you run `compile_ios_tensorflow`.
|
||||
(These should be called by `build_all_ios.sh` if you are using it, but check
|
||||
if they have run successful.)
|
||||
|
||||
- Check that you have version 7.3 of Xcode.
|
||||
|
||||
- If there are Eigen errors, look inside the build settings of your Xcode project. In the Search Paths section, you'll see an Eigen include directory that changes with each version of the framework. You may need to update this to may the version in your tensorflow/contrib/makefile/downloads folder.
|
||||
|
||||
- If there's a complaint about no Session's registered, that means that the C++ global constructors that TensorFlow relies on for registration haven't been linked in properly. You'll have to make sure your project uses force_load, as described below.
|
||||
- If there's a complaint about no Sessions registered, that means that the C++
|
||||
global constructors that TensorFlow relies on for registration haven't been
|
||||
linked in properly. You'll have to make sure your project uses force_load, as
|
||||
described below.
|
||||
|
||||
## Creating your Own App
|
||||
|
||||
You'll need to update various settings in your app to link against TensorFlow. You can view them in the example projects, but here's a full rundown:
|
||||
You'll need to update various settings in your app to link against
|
||||
TensorFlow. You can view them in the example projects, but here's a full
|
||||
rundown:
|
||||
|
||||
- The `compile_ios_tensorflow.sh' script builds a universal static library in tensorflow/contrib/makefile/gen/lib/libtensorflow-core.a. You'll need to add this to your linking build stage, and in Search Paths add tensorflow/contrib/makefile/gen/lib to the Library Search Paths setting.
|
||||
- The `compile_ios_tensorflow.sh` script builds a universal static library in
|
||||
`tensorflow/contrib/makefile/gen/lib/libtensorflow-core.a`. You'll need to add
|
||||
this to your linking build stage, and in Search Paths add
|
||||
`tensorflow/contrib/makefile/gen/lib` to the Library Search Paths setting.
|
||||
|
||||
- You'll also need to add libprotobuf.a and libprotobuf-lite.a from tensorflow/contrib/makefile/gen/protobuf_ios/lib to your Build Stages and Library Search Paths.
|
||||
- You'll also need to add `libprotobuf.a` and `libprotobuf-lite.a` from
|
||||
`tensorflow/contrib/makefile/gen/protobuf_ios/lib` to your _Build Stages_ and
|
||||
_Library Search Paths_.
|
||||
|
||||
- The Header Search paths needs to contain the root folder of tensorflow, tensorflow/contrib/makefile/downloads/protobuf/src, tensorflow/contrib/makefile/downloads, tensorflow/contrib/makefile/downloads/eigen-eigen-<current Eigen hash>, and tensorflow/contrib/makefile/gen/proto.
|
||||
- The _Header Search_ paths needs to contain:
|
||||
- the root folder of tensorflow,
|
||||
- `tensorflow/contrib/makefile/downloads/protobuf/src`
|
||||
- `tensorflow/contrib/makefile/downloads`,
|
||||
- `tensorflow/contrib/makefile/downloads/eigen-latest`, and
|
||||
- `tensorflow/contrib/makefile/gen/proto`.
|
||||
|
||||
- In the Linking section, you need to add `-force_load` followed by the path to the TensorFlow static library in the Other Linker Flags section. This ensures that the global C++ objects that are used to register important classes inside the library are not stripped out. To the linker, they can appear unused because no other code references the variables, but in fact their constructors have the important side effect of registering the class.
|
||||
- In the Linking section, you need to add `-force_load` followed by the path to
|
||||
the TensorFlow static library in the _Other Linker_ Flags section. This ensures
|
||||
that the global C++ objects that are used to register important classes
|
||||
inside the library are not stripped out. To the linker, they can appear
|
||||
unused because no other code references the variables, but in fact their
|
||||
constructors have the important side effect of registering the class.
|
||||
|
||||
- The library doesn't currently support bitcode, so you'll need to disable that in your project settings.
|
||||
- The library doesn't currently support bitcode, so you'll need to disable that
|
||||
in your project settings.
|
||||
|
||||
@ -1,34 +1,81 @@
|
||||
### TensorFlow Makefile
|
||||
|
||||
The recommended way to build TensorFlow from source is using the Bazel
|
||||
open-source build system. Sometimes this isn't possible though:
|
||||
open-source build system. Sometimes this isn't possible.
|
||||
|
||||
- The system may not have the RAM or processing power to support Bazel.
|
||||
- Bazel dependencies might not be available.
|
||||
- The build system may not have the RAM or processing power to support Bazel.
|
||||
- Bazel or its dependencies may not be available.
|
||||
- You may want to cross-compile for an unsupported target system.
|
||||
|
||||
This experimental project supplies a Makefile automatically derived from the
|
||||
dependencies listed in the Bazel project, that can be used with GNU's make tool.
|
||||
It offers the ability to compile the core C++ runtime into a static library, but
|
||||
doesn't include more advanced features like Python or other language bindings,
|
||||
or GPU support.
|
||||
dependencies listed in the Bazel project that can be used with GNU's make tool.
|
||||
With it, you can compile the core C++ runtime into a static library.
|
||||
|
||||
## Building
|
||||
This static library will not contain:
|
||||
|
||||
To compile the library and an example program using it, first pull the
|
||||
dependencies:
|
||||
- Python or other language bindings
|
||||
- GPU support
|
||||
|
||||
You can target:
|
||||
- iOS
|
||||
- OS X (macOS)
|
||||
- Android
|
||||
- Raspberry-PI
|
||||
|
||||
You will compile tensorflow and protobuf libraries that you can link into other
|
||||
applications. You will also compile the [benchmark](../../tools/benchmark/)
|
||||
application that will let you check your application.
|
||||
|
||||
## Before you start (all platforms)
|
||||
|
||||
```bash
|
||||
First, clone this TensorFlow repository.
|
||||
|
||||
You will need to download all dependencies as well. We have provided a script
|
||||
that does so, to be run (as with all commands) at the root of the repository:
|
||||
|
||||
```bash
|
||||
tensorflow/contrib/makefile/download_dependencies.sh
|
||||
```
|
||||
|
||||
You should only need to do this step once, it puts required libraries like Eigen
|
||||
in the `tensorflow/contrib/makefile/downloads/` folder. You will also need to
|
||||
make sure you have a version of [protobuf 3](https://github.com/google/protobuf)
|
||||
installed on your system, either through package management or building from
|
||||
source.
|
||||
You should only need to do this step once. It downloads the required libraries
|
||||
like Eigen in the `tensorflow/contrib/makefile/downloads/` folder.
|
||||
|
||||
Then you can build the project:
|
||||
You should download the example graph from [https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip](https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip).
|
||||
|
||||
## Building on Linux
|
||||
|
||||
_Note: This has only been tested on Ubuntu._
|
||||
|
||||
Don't forget to download dependencies if you haven't already:
|
||||
|
||||
```bash
|
||||
tensorflow/contrib/makefile/download_dependencies.sh
|
||||
```
|
||||
|
||||
You will need install a version of
|
||||
[protobuf 3](https://github.com/google/protobuf) on your system. We strongly
|
||||
recommend that you compile and install the version downloaded in the script
|
||||
above.
|
||||
|
||||
On Ubuntu, you can do this:
|
||||
```bash
|
||||
sudo apt-get install autoconf automake libtool curl make g++ unzip
|
||||
pushd .
|
||||
cd tensforflow/contrib/makefile/downloads/protobuf
|
||||
./autogen.sh
|
||||
./configure
|
||||
make
|
||||
make check
|
||||
sudo make install
|
||||
sudo ldconfig # refresh shared library cache
|
||||
popd
|
||||
```
|
||||
|
||||
If you have issues (or can't use apt-get), see
|
||||
[these instructions](https://github.com/google/protobuf/blob/master/src/README.md)
|
||||
for specific installation of C++ support tools.
|
||||
|
||||
After you have installed protobufs, you can run this from the repository root:
|
||||
|
||||
```bash
|
||||
make -f tensorflow/contrib/makefile/Makefile
|
||||
@ -36,100 +83,166 @@ make -f tensorflow/contrib/makefile/Makefile
|
||||
|
||||
This should compile a static library in
|
||||
`tensorflow/contrib/makefile/gen/lib/tf_lib.a`, and create an example executable
|
||||
at `tensorflow/contrib/makefile/gen/bin/benchmark`. To run the executable, use:
|
||||
at `tensorflow/contrib/makefile/gen/bin/benchmark`.
|
||||
|
||||
Get the graph file, if you have not already:
|
||||
|
||||
```bash
|
||||
tensorflow/contrib/makefile/gen/bin/benchmark --graph=tensorflow_inception_graph.pb
|
||||
mkdir -p ~/graphs
|
||||
curl -o ~/graphs/inception.zip \
|
||||
https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip \
|
||||
&& unzip ~/graphs/inception.zip -d ~/graphs/inception
|
||||
```
|
||||
|
||||
You should download the example graph from [https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip](https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip).
|
||||
To run the executable, use:
|
||||
|
||||
## Supported Systems
|
||||
|
||||
The script has been tested on Ubuntu, OS X, Android, and iOS. If you look in the
|
||||
Makefile itself, you'll see it's broken up into host and target sections. If you
|
||||
are cross-compiling, you should look at customizing the target settings to match
|
||||
what you need for the system you're aiming at.
|
||||
```bash
|
||||
tensorflow/contrib/makefile/gen/bin/benchmark \
|
||||
--graph=~/graphs/inception/tensorflow_inception_graph.pb
|
||||
```
|
||||
|
||||
## Android
|
||||
|
||||
For Android, you'll need to explicitly specify that as the target, and supply
|
||||
the location of the NDK toolchain on the command line, for example:
|
||||
First, you will need to download and unzip the
|
||||
[Native Development Kit (NDK)](http://developers.google.com/ndk). You will not
|
||||
need to install the standalone toolchain, however.
|
||||
|
||||
Assign your NDK location to $NDK_ROOT:
|
||||
|
||||
```bash
|
||||
make -f tensorflow/contrib/makefile/Makefile \
|
||||
TARGET=ANDROID \
|
||||
ANDROID_NDK_DIR=$(HOME)/toolchains/clang-21-stl-gnu
|
||||
export NDK_ROOT=/absolute/path/to/NDK/android-ndk-rxxx/
|
||||
```
|
||||
|
||||
You'll also need a compiled version of the protobuf libraries for Android. You
|
||||
can use the helper script at `compile_android_protobuf.sh` to create these.
|
||||
Download the graph if you haven't already:
|
||||
|
||||
```bash
|
||||
mkdir -p ~/graphs
|
||||
curl -o ~/graphs/inception.zip \
|
||||
https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip \
|
||||
&& unzip ~/graphs/inception.zip -d ~/graphs/inception
|
||||
```
|
||||
|
||||
Then, execute the following:
|
||||
|
||||
```bash
|
||||
tensorflow/contrib/makefile/download_dependencies.sh
|
||||
tensorflow/contrib/makefile/compile_android_protobuf.sh -c
|
||||
make -f tensorflow/contrib/makefile/Makefile TARGET=ANDROID
|
||||
```
|
||||
|
||||
At this point, you will have compiled libraries in `gen/lib/*` and the
|
||||
[benchmark app](../../tools/benchmark) compiled for Android.
|
||||
|
||||
Run the benchmark by pushing both the benchmark and the graph file to your
|
||||
attached Android device:
|
||||
|
||||
```bash
|
||||
adb push ~/graphs/inception/tensorflow_inception_graph.pb /data/local/tmp/
|
||||
adb push tensorflow/contrib/makefile/gen/bin/benchmark /data/local/tmp/
|
||||
adb shell '/data/local/tmp/benchmark \
|
||||
--graph=/data/local/tmp/classify_image_graph_def.pb \
|
||||
--input_layer="input:0" \
|
||||
--input_layer_shape="1,224,224,3" \
|
||||
--input_layer_type="float" \
|
||||
--output_layer="output:0"
|
||||
'
|
||||
```
|
||||
|
||||
For more details, see the [benchmark documentation](../../tools/benchmark).
|
||||
|
||||
## iOS
|
||||
|
||||
For iOS you'll need to first run `compile_ios_protobuf.sh` to build iOS versions
|
||||
of the protobuf libraries. Once that's complete, you can run the makefile
|
||||
specifying iOS as the target, along with the architecture you want to build for:
|
||||
_Note: To use this library in an iOS application, see related instructions in
|
||||
the [iOS examples](../ios_examples/] directory._
|
||||
|
||||
Install XCode 7.3 or more recent. If you have not already, you will need to
|
||||
install the command-line tools using `xcode-select`:
|
||||
|
||||
```bash
|
||||
make -f tensorflow/contrib/makefile/Makefile \
|
||||
TARGET=IOS \
|
||||
IOS_ARCH=ARM64
|
||||
xcode-select --install
|
||||
```
|
||||
|
||||
This will build the library for a single architecture, and the benchmark
|
||||
program. Since the benchmark is command-line only, you'll need to load the
|
||||
static library into an Xcode app project to use it.
|
||||
If this is a new install, you will need to run XCode once to agree to the
|
||||
license before continuing.
|
||||
|
||||
To build a complete universal library for iOS, containing all architectures,
|
||||
you will need to run `compile_ios_protobuf.sh` followed by
|
||||
`compile_ios_tensorflow.sh`. This creates a library in
|
||||
`tensorflow/contrib/makefile/gen/lib/libtensorflow-core.a` that you can link any
|
||||
xcode project against. Here are complete build instructions:
|
||||
|
||||
Grab the source code for TensorFlow:
|
||||
Also, download the graph if you haven't already:
|
||||
|
||||
```bash
|
||||
git clone https://github.com/tensorflow/tensorflow
|
||||
cd tensorflow
|
||||
mkdir -p ~/graphs
|
||||
curl -o ~/graphs/inception.zip \
|
||||
https://storage.googleapis.com/download.tensorflow.org/models/inception5h.zip \
|
||||
&& unzip ~/graphs/inception.zip -d ~/graphs/inception
|
||||
```
|
||||
|
||||
Download dependencies like Eigen and Protobuf:
|
||||
### Building all at once
|
||||
|
||||
If you just want to get the libraries compiled in a hurry, you can run:
|
||||
|
||||
```bash
|
||||
build_all_ios.sh
|
||||
```
|
||||
|
||||
and wait a long time.
|
||||
|
||||
When this completes, you will have a library for a single architecture and the
|
||||
benchmark program. Although successfully compiling the benchmark program is a
|
||||
sign of success, the program is not a complete iOS app.
|
||||
|
||||
To see TensorFlow running on iOS, the example Xcode project in
|
||||
[tensorflow/contrib/ios_example](../ios_example) shows how to use the static
|
||||
library in a simple app.
|
||||
|
||||
### Building by hand
|
||||
|
||||
This section covers each step of building. For all the code in one place, see
|
||||
[build_all_ios.sh](build_all_ios.sh).
|
||||
|
||||
If you have not already, you will need to download dependencies:
|
||||
|
||||
```bash
|
||||
tensorflow/contrib/makefile/download_dependencies.sh
|
||||
```
|
||||
|
||||
Build and install the host (OS X) copy of protobuf.
|
||||
Next, you will need to compile protobufs for iOS:
|
||||
|
||||
```bash
|
||||
cd tensorflow/contrib/makefile/downloads/protobuf/
|
||||
./autogen.sh
|
||||
./configure
|
||||
make
|
||||
sudo make install
|
||||
cd ../../../../..
|
||||
compile_ios_protobuf.sh
|
||||
```
|
||||
|
||||
Build the iOS native versions of protobuf:
|
||||
Then, you can run the makefile specifying iOS as the target, along with the
|
||||
architecture you want to build for:
|
||||
|
||||
```bash
|
||||
tensorflow/contrib/makefile/compile_ios_protobuf.sh
|
||||
make -f tensorflow/contrib/makefile/Makefile \
|
||||
TARGET=IOS \
|
||||
IOS_ARCH=ARM64
|
||||
```
|
||||
|
||||
Build all iOS architectures for TensorFlow:
|
||||
This creates a library in
|
||||
`tensorflow/contrib/makefile/gen/lib/libtensorflow-core.a` that you can link any
|
||||
xcode project against.
|
||||
|
||||
At this point, you will have a library for a single architecture and the
|
||||
benchmark program. Although successfully compiling the benchmark program is a
|
||||
sign of success, the program is not a complete iOS app.
|
||||
|
||||
To see TensorFlow running on iOS, the example Xcode project in
|
||||
[tensorflow/contrib/ios_example](../ios_example) shows how to use the static
|
||||
library in a simple app.
|
||||
|
||||
#### Universal binaries
|
||||
|
||||
In some situations, you will need a universal library. In that case, you will
|
||||
still need to run `compile_ios_protobuf.sh`, but this time follow it with:
|
||||
|
||||
```bash
|
||||
tensorflow/contrib/makefile/compile_ios_tensorflow.sh
|
||||
compile_ios_tensorflow.sh
|
||||
```
|
||||
|
||||
You will need to use -force_load in the linker flags
|
||||
In XCode, you will need to use -force_load in the linker flags
|
||||
section of the build settings to pull in the global constructors that are used
|
||||
to register ops and kernels.
|
||||
|
||||
The example Xcode project in tensorflow/contrib/ios_example shows how to use the
|
||||
static library in a simple app.
|
||||
|
||||
## Raspberry Pi
|
||||
|
||||
Building on the Raspberry Pi is similar to a normal Linux system, though we
|
||||
@ -155,10 +268,19 @@ some extra optimization flags to give you code that will run faster:
|
||||
|
||||
```bash
|
||||
make -f tensorflow/contrib/makefile/Makefile HOST_OS=PI TARGET=PI \
|
||||
OPTFLAGS="-Os -mfpu=neon-vfpv4 -funsafe-math-optimizations -ftree-vectorize"
|
||||
OPTFLAGS="-Os -mfpu=neon-vfpv4 -funsafe-math-optimizations -ftree-vectorize"
|
||||
```
|
||||
|
||||
## Dependencies
|
||||
# Other notes
|
||||
|
||||
## Supported Systems
|
||||
|
||||
The Make script has been tested on Ubuntu and OS X. If you look in the Makefile
|
||||
itself, you'll see it's broken up into host and target sections. If you are
|
||||
cross-compiling, you should look at customizing the target settings to match
|
||||
what you need for your desired system.
|
||||
|
||||
## Dependency Managment
|
||||
|
||||
The Makefile loads in a list of dependencies stored in text files. These files
|
||||
are generated from the main Bazel build by running
|
||||
@ -171,3 +293,16 @@ make header changes you will need to run this command to recompile cleanly:
|
||||
```bash
|
||||
make -f tensorflow/contrib/makefile/Makefile clean
|
||||
```
|
||||
|
||||
### Cleaning up
|
||||
|
||||
In some situations, you may want to completely clean up. The dependencies,
|
||||
intermediate stages, and generated files are stored in:
|
||||
|
||||
```bash
|
||||
tensorflow/contrib/makefile/downloads
|
||||
tensorflow/contrib/makefile/gen
|
||||
```
|
||||
|
||||
Those directories can safely be removed, but you will have to start over with
|
||||
`download_dependencies.sh` once you delete them.
|
||||
|
||||
Loading…
Reference in New Issue
Block a user