TensorFlow for Mobile Developers

Transcript

1. 1.

   TensorFlow for Mobile Developers Enrique López Mañas Google Developer Expert

2. 2.

   Ego Slide • Freelance Dev • Google Developer Expert •

   @eenriquelopez
3. 3.

   Machine Learning / AI

4. 4.

   Machine Learning / AI

5. 5.

   Machine Learning / AI

6. 6.

   Buzzwords classification Artificial intelligence Machine Learning The science of making

   machines smarts Building machines that can learn Neural Networks One of the many different algorithms in Machine Learning
7. 7.

   Buzzwords classification

8. 8.

   Neural Network

9. 9.

   Machine Learning

10. 10.

    Machine Learning

11. 11.

    Machine Learning 12% of all responses on mobile

12. 12.

    Machine Learning

13. 13.

    Machine Learning

14. 14.

    Machine Learning

15. 15.

    Cloud vs. Mobile Less Traffic & Faster Response Motion Sensors

16. 16.

    TensorFlow OpenSource library for Machine Learning tensorflow.org Today most popular

    ML framework
17. 17.

    TensorFlow You can train: - Mac / Windows - GPU

    Server - GPU/TPU on Cloud Prediction: - Android - iOS - Raspberry
18. 18.

    TensorFlow CIFAR10

19. 19.

    TensorFlow ecosystem

20. 20.

    From Training to App (@yufengg)

21. 21.

    Data Gathering?

22. 22.

    From Training to App (@yufengg)

23. 23.

    From Training to App (@yufengg)

24. 24.

    Convolutional Neural Networks (@yufengg)

25. 25.

    Convolutional Neural Networks (@yufengg)

26. 26.

    Convolutional Neural Networks (@yufengg)

27. 27.

    Convolutional Neural Networks (@yufengg)

28. 28.

    Convolutional Neural Networks (@yufengg)

29. 29.

    Convolutional Neural Networks (@yufengg)

30. 30.

    Convolutional Neural Networks (@yufengg)

31. 31.

    Convolutional Neural Networks (@yufengg)

32. 32.

    From Training to App (@yufengg)

33. 33.

    Optimizing model for Mobile (@yufengg)

34. 34.

    From Training to App (@yufengg) Other inception versions (inception v1

    quantised is 7 MB)
35. 35.

    From Training to App (@yufengg) TensorFlow increases APK in 12

    MB
36. 36.

    From Training to App (@yufengg)

37. 37.

    From Training to App (@yufengg)

38. 38.

    TensorFlow community 1000+ contributors 22.000+ commits 18.000+ repositories with name

    “TensorFlow”
39. 39.

    TensorFlow community

40. 40.

    Building with Android Studio Uses Bazel to build Set Bazel

    binary location in /tensorflow/examples/ android/build.gradle Add project in tensorflor/examples/android folder to Android Studio
41. 41.

    Building with Android Studio Installing Bazel first: Download from https://github.com/bazelbuild/bazel/releases

    Install instructions in https://docs.bazel.build/versions/master/install.html
42. 42.

    Building with Android Studio Download Android SDK (23) Download Android

    NDK (12b)
 
 (recommend it with Android SDK Manager)
43. 43.

    Building with Android Studio WORKSPACE FILE

44. 44.

    Building with Android Studio Run Bazel bazel build -c opt

    //tensorflow/examples/ android:tensorflow_demo
45. 45.

    Building with Android Studio Install APK adb install -r bazel-bin/tensorflow/examples/android/

    tensorflow_demo.apk
46. 46.

    Building with Android Studio OR… TensorFlow AAR from JCenter YUV

    -> RGB less efficient Object tracking not available
47. 47.

    Android Samples TF Classify 
 TF Detect
 
 TF Stylize

48. 48.

    TF Classify Uses Google Inception (v3) to label images Model

    easy to swap No “person label” Volume button up for statistics
49. 49.

    TF Classify

50. 50.

    TF Classify

51. 51.

    TF Detect

52. 52.

    TF Detect - Draw bounding boxes around people - Useful

    to count objects - No training yet
53. 53.

    TF Style

54. 54.

    TF Style - Real time style transfer algorithm - Pick/mix

    different styles - Can train Magenta models
55. 55.

    TF Style

56. 56.

    TensorFlow Written in C++ Android uses Kotlin/Java?
 
 HowTo? Android

    inference Library https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/android
57. 57.

    Building for iOS Install requirements: - Xcode 8 - Command

    line tools (xcode-select install) - brew install automake - brew install lib tool - tensorflow/contrib/makefile_build_all_ios.sh (takes 20 minutes)
58. 58.

    iOS Examples: simple

59. 59.

    iOS Examples: camera

60. 60.

    iOS Examples: camera - Run inception each frame - Models

    can be replaced
61. 61.

    iOS Examples: benchmark

62. 62.

    Raspberry Pi

63. 63.

    Resources TensorFlow: https://www.tensorflow.org/ Magenta: https://magenta.tensorflow.org/welcome-to-magenta TensorFlow for Poets: https://codelabs.developers.google.com/codelabs/ tensorflow-for-poets/index.html#0

64. 64.

    Feedback http://bit.ly/Mobius2017

65. 65.

    None