Skip to content

use 'official' TensorFlow lib #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 7 commits into
base: master
Choose a base branch
from
Open

use 'official' TensorFlow lib #3

Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
fd95f87
clean up Android project
Piasy May 7, 2017
0d16120
use nightly build tensorflow artifact of http://ci.tensorflow.org/vie…
Piasy May 7, 2017
89b3354
update README
Piasy May 7, 2017
a3383e2
Use official tensorflow-android release
Piasy May 28, 2017
4b7ae9a
rebuild model, simplify model export way
Piasy May 29, 2017
0a54fc8
rearrange python code
Piasy May 29, 2017
c1008fb
include model trained by Keras
Piasy Jun 1, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
update README
  • Loading branch information
@Piasy
Piasy committed May 7, 2017
commit 89b33543d5749c5857ad46b1b590474729f3d77c
20 changes: 2 additions & 18 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,9 +12,6 @@ how to save your model and export it for Android or other devices check the
very simple tutorial bellow.

The UI and expert-graph.pb model were taken from: https://github.com/miyosuda/TensorFlowAndroidMNIST, so thank you miyousuda.
The TensorFlow jar and so armeabi-v7a were taken from: https://github.com/MindorksOpenSource/AndroidTensorFlowMNISTExample,
so thank you MindorksOpenSource.
The Tensorflow so of x86 was taken from: https://github.com/cesardelgadof/TensorFlowAndroidMNIST, so thank you cesardelgadof.

If you have no ideia what I just said above, have a look on the instructions bellow.

Expand All @@ -32,7 +29,7 @@ A full example can be seen [here](https://github.com/mari-linhares/mnist-android
Example: `_w = sess.eval(w)`, where w was learned from training.
3. Rewrite your model changing the variables for constants with value = in memory copy of learned variables.
Example: `w_save = tf.constant(_w)`

Also make sure to put names in the input and output of the model, this will be needed for the model later.
Example:
`x = tf.placeholder(tf.float32, [None, 1000], name='input')`
Expand All @@ -42,23 +39,10 @@ A full example can be seen [here](https://github.com/mari-linhares/mnist-android

## How to run my model with Android?

You need two things:

1. [The TensorFlow jar](https://github.com/MindorksOpenSource/AndroidTensorFlowMNISTExample/blob/master/app/libs/libandroid_tensorflow_inference_java.jar)
Move it to the libs folder, right click and add as library.

2. The TensorFlow so file for the desired architecture:
[x86](https://github.com/cesardelgadof/TensorFlowAndroidMNIST/blob/master/app/src/main/jniLibs/x86/libtensorflow_mnist.so)
[armeabi-v7a](https://github.com/MindorksOpenSource/AndroidTensorFlowMNISTExample/tree/master/app/src/main/jniLibs/armeabi-v7a)

Creat the jniLibs/x86 folder or the jniLibs/armeabi-v7a folder at the main folder.
Move it to app/src/main/jniLibs/x86/libtensorflow_inference.so or app/src/jniLibs/armeabi-v7a/libtensorflow_inference.so

If you want to generate these files yourself, [here](https://blog.mindorks.com/android-tensorflow-machine-learning-example-ff0e9b2654cc) is a nice tutorial of how to do it.
You need `tensorflow.aar`, which can be downloaded from [the nightly build artifact of TensorFlow CI](http://ci.tensorflow.org/view/Nightly/job/nightly-android/), here we use [the #124 build](http://ci.tensorflow.org/view/Nightly/job/nightly-android/124/artifact/).

## Interacting with TensorFlow

To interact with TensorFlow you will need an instance of TensorFlowInferenceInterface, you can see more details about it [here](https://github.com/mari-linhares/mnist-android-tensorflow/blob/master/MnistAndroid/app/src/main/java/mariannelinhares/mnistandroid/Classifier.java)

Thank you, have fun!