TensorFlow for Mobile Developers

TensorFlow for Mobile Developers @ DroidKaigi

Transcript

1. TensorFlow for
   Mobile Developers
   Enrique López Mañas
   Google Developer Expert
   

   View Slide

2. Ego Slide
   • Freelance Dev
   • Google Developer Expert
   • @eenriquelopez
   

   View Slide

3. Machine Learning / AI
   

   View Slide

4. Machine Learning / AI
   

   View Slide

5. Machine Learning / AI
   

   View Slide

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
   

   View Slide

7. Buzzwords classification
   

   View Slide

8. Neural Network
   

   View Slide

9. Machine Learning
   

   View Slide

10. Machine Learning
    

    View Slide

11. Machine Learning
    12%
    of all
    responses
    on mobile
    

    View Slide

12. Machine Learning
    

    View Slide

13. Machine Learning
    

    View Slide

14. Machine Learning
    

    View Slide

15. Cloud vs. Mobile
    Less Traffic & Faster Response
    Motion Sensors
    

    View Slide

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

    View Slide

17. TensorFlow
    You can train:
    - Mac / Windows
    - GPU Server
    - GPU/TPU on Cloud
    Prediction:
    - Android
    - iOS
    - Raspberry
    

    View Slide

18. TensorFlow
    CIFAR10
    

    View Slide

19. TensorFlow ecosystem
    

    View Slide

20. From Training to App (@yufengg)
    

    View Slide

21. Data Gathering?
    

    View Slide

22. From Training to App (@yufengg)
    

    View Slide

23. From Training to App (@yufengg)
    

    View Slide

24. Convolutional Neural Networks (@yufengg)
    

    View Slide

25. Convolutional Neural Networks (@yufengg)
    

    View Slide

26. Convolutional Neural Networks (@yufengg)
    

    View Slide

27. Convolutional Neural Networks (@yufengg)
    

    View Slide

28. Convolutional Neural Networks (@yufengg)
    

    View Slide

29. Convolutional Neural Networks (@yufengg)
    

    View Slide

30. Convolutional Neural Networks (@yufengg)
    

    View Slide

31. Convolutional Neural Networks (@yufengg)
    

    View Slide

32. From Training to App (@yufengg)
    

    View Slide

33. Optimizing model for Mobile (@yufengg)
    

    View Slide

34. From Training to App (@yufengg)
    Other inception versions
    (inception v1 quantised is 7 MB)
    

    View Slide

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

    View Slide

36. From Training to App (@yufengg)
    

    View Slide

37. From Training to App (@yufengg)
    

    View Slide

38. TensorFlow community
    1000+ contributors
    22.000+ commits
    18.000+ repositories with name “TensorFlow”
    

    View Slide

39. TensorFlow community
    

    View Slide

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
    

    View Slide

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
    

    View Slide

42. Building with Android Studio
    Download Android SDK (23)
    Download Android NDK (12b)

    

    (recommend it with Android SDK Manager)
    

    View Slide

43. Building with Android Studio
    WORKSPACE FILE
    

    View Slide

44. Building with Android Studio
    Run Bazel
    bazel build -c opt //tensorflow/examples/
    android:tensorflow_demo
    

    View Slide

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

    View Slide

46. Building with Android Studio
    OR…
    TensorFlow AAR from JCenter
    YUV -> RGB less efficient
    Object tracking not available
    

    View Slide

47. Android Samples
    TF Classify
    

    TF Detect

    

    TF Stylize
    

    View Slide

48. TF Classify
    Uses Google Inception (v3) to label images
    Model easy to swap
    No “person label”
    Volume button up for statistics
    

    View Slide

49. TF Classify
    

    View Slide

50. TF Classify
    

    View Slide

51. TF Detect
    

    View Slide

52. TF Detect
    - Draw bounding boxes around people
    - Useful to count objects
    - No training yet
    

    View Slide

53. TF Style
    

    View Slide

54. TF Style
    - Real time style transfer algorithm
    - Pick/mix different styles
    - Can train Magenta models
    

    View Slide

55. TF Style
    

    View Slide

56. TensorFlow
    Written in C++
    Android uses Kotlin/Java?

    

    HowTo?
    Android inference Library
    https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/android
    

    View Slide

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)
    

    View Slide

58. iOS Examples: simple
    

    View Slide

59. iOS Examples: camera
    

    View Slide

60. iOS Examples: camera
    - Run inception each frame
    - Models can be replaced
    

    View Slide

61. iOS Examples: benchmark
    

    View Slide

62. Raspberry Pi
    

    View Slide

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
    

    View Slide

64. Shameless self-promotion
    www.kotlinweekly.net
    

    View Slide

65. Feedback!
    http://bit.ly/droidkaigi
    

    View Slide

66. View Slide