DevFest Tokyo 2017 TensorFlow

DevFest Tokyo 2017で発表した「TensorFlowで
趣味の画像収集サーバーを作る 2017年10月号」の資料です。

SSD（Single Shot MultiBox Detector）によるイラスト顔検出について。

Transcript

1. C-LIS CO., LTD.
   

   View Slide

2. C-LIS CO., LTD.
   ༗ࢁܓೋʢ,FJKJ"3*:"."ʣ
   $-*4$0 -5%
   Photo : Koji MORIGUCHI (AUN CREATIVE FIRM)
   "OESPJEΞϓϦ։ൃνϣοτσΩϧ
   ػցֶश͸ͪΐͬͱ΍ͬͨ͜ͱ͋Γ·͢
   ΍ͬͯ·ͤΜ
   

   View Slide

3. %FW'FTU5PLZP
   
   5FOTPS'MPXͰ

   झຯͷը૾ऩूαʔόʔΛ࡞Δ
   ೥݄߸
   

   View Slide

4. 
   લճ·Ͱͷ͓࿩
   

   View Slide

5. 5FOTPS'MPXൃදʢ೥݄ʣ
   
   

   View Slide

6. 
   ษڧձ΍Ζ͏ͥ
   

   View Slide

7. 
   (PPHMF%FWFMPQFS(SPVQ
   

   View Slide

8. 
   IUUQTHEHLPCFEPPSLFFQFSKQFWFOUT
   

   View Slide

9. Πϯλʔωοτ͔Β
   ޷Έͷը૾ΛࣗಈͰऩू͍ͨ͠
   

   View Slide

10. © ࠜઇΕ͍
    ؟
    ڸ
    ͬ
    ່
    

    View Slide

11. γεςϜߏ੒
    
    Downloader Face Detection Megane Detection
    ֬ೝɾमਖ਼
    ೝࣝ݁Ռ
    JSON
    ֶशʢ܇࿅ʣ
    λΠϜϥΠϯ ϝσΟΞ
    σʔληοτ
    ֶशʢ܇࿅ʣ
    TensorFlow
    
    

    View Slide

12. {
    "generator": "Region Cropper",
    "file_name": "haruki_g17.png",
    "regions": [
    {
    "probability": 1.0,
    "label": 2,
    "rect": {
    "left": 97.0,
    "top": 251.0,
    "right": 285.0,
    "bottom": 383.0
    }
    },
    {
    "probability": 1.0,
    "label": 2,
    "rect": {
    "left": 536.0,
    "top": 175.0,
    "right": 730.0,
    "bottom": 321.0
    }
    }
    ]
    }
    Region Cropper: https://github.com/keiji/region_cropper
    
    

    View Slide

13. ؟ڸ່ͬ൑ఆ
    conv
    3x3x3
    32
    conv
    3x3x32
    64
    max_pooling lrn conv
    3x3x64

    128
    conv
    3x3x128
    128
    max_pooling
    lrn conv
    3x3x128

    128
    conv
    3x3x128
    256
    fc
    384
    fc
    256
    output
    2
    
    

    View Slide

14. ͷ൑ఆਫ਼౓
    1 0
    

    View Slide

15. ٕज़ܥಉਓࢽΦϯϦʔΠϕϯτ
    ٕज़ॻయ̍ˍ̎
    
    https://techbookfest.org/tbf01/#A-28 https://techbookfest.org/event/tbf02/circle/5705718560718848
    

    View Slide

16. ੔ཧ଴ͪͷը૾
    ຕ
    
    ʢ2017.10.08 ࣌఺ʣ
    

    View Slide

17. σʔληοτͷ੔උʢ੔ཧʣ͕

    ௚ۙͷ՝୊
    

    View Slide

18. Πϥετإݕग़ʢ'BDF%FUFDUJPOʣ
    
    

    View Slide

19. γεςϜߏ੒
    
    Downloader Face Detection Megane Detection
    ֬ೝɾमਖ਼
    ೝࣝ݁Ռ
    JSON
    ֶशʢ܇࿅ʣ
    λΠϜϥΠϯ ϝσΟΞ
    σʔληοτ
    ֶशʢ܇࿅ʣ
    TensorFlow
    
    

    View Slide

20. 
    3$//
    3FHJPOCBTFE$POWPMVUJPOBM/FVSBM/FUXPSLT
    

    View Slide

21. 4MJEJOH8JOEPX
    
    

    View Slide

22. 4FMFDUJWF4FBSDI
    
    4FMFDUJWF4FBSDIGPS0CKFDU3FDPHOJUJPO
    IUUQXXXIVQQFMFOOMQVCMJDBUJPOTTFMFDUJWF4FBSDI%SBGUQEG
    848
    

    View Slide

23. %FW'FTU5PLZP
    
    
    5FOTPS'MPXͰ
    Πϥετإݕग़ثΛ࡞Δ
    

    View Slide

24. ֶशσʔλͷ࡞੒
    

    View Slide

25. ࠲ඪͷਖ਼ଇԽɹਖ਼نԽ
    888x613
    left : 536 → 0.6036
    top : 175 → 0.2854
    right : 730 → 0.8220
    bottom : 321 → 0.5236
    0.1092 ← 97 : left
    0.4094 ← 251 : top
    0.3209 ← 285 : right
    0.6247 ← 383 : bottom
    
    

    View Slide

26. ը૾ͷϦαΠζɾάϨʔεέʔϧԽ
    256x256
    left : 0.6036
    top : 0.2854
    right : 0.8220
    bottom : 0.5236
    0.1092 : left
    0.4094 : top
    0.3209 : right
    0.6247 : bottom
    
    

    View Slide

27. Ϩίʔυߏ଄
    ը૾σʔλʢόΠτྻʣ
    
    

    

    

    
    
    

    View Slide

28. XJUIUGQZUIPO@JP5'3FDPSE8SJUFS PVUQVU@QBUI PQUJPOT
    BTXSJUFS
    FYBNQMFUGUSBJO&YBNQMF GFBUVSFTUGUSBJO'FBUVSFT GFBUVSF\
    GJMF@OBNF@CZUFT@GFBUVSF GJMF@OBNFFODPEF
    
    
    JNBHF@CZUFT@GFBUVSF JNBHF@EBUB
    
    SFHJPO@DPVOU@JOU@GFBUVSF SFHJPO@DPVOU
    
    SFHJPOT@GMPBU@GFBUVSF GMBUUFO@SFHJPOT
    
    MBCFMT@JOU@MJTU@GFBUVSF MBCFMT
    
    ^
    
    
    XSJUFSXSJUF FYBNQMF4FSJBMJ[F5P4USJOH
    
    
    
    

    View Slide

29. Ϟσϧ
    conv
    5x5x1

    128
    batch_normalization
    conv
    1x1x128

    64
    conv
    3x3x64

    64
    batch_normalization
    conv
    1x1x64

    128
    ReLU
    conv
    1x1x128

    128
    conv
    3x3x128

    128
    batch_normalization
    conv
    1x1x128

    128
    ReLU
    conv
    1x1x256

    256
    conv
    3x3x256

    256
    batch_normalization
    conv
    1x1x256

    128
    ReLU
    conv
    2x2x128

    128
    fc
    2048
    fc
    4
    256x256x1
    Residual block 1 Residual block 2 Residual block 3
    × 3
    × 3
    × 3
    avg_pool
    2x2
    avg_pool
    2x2
    avg_pool
    2x2
    ReLU Sigmoid
    
    

    View Slide

30. ೖྗͱग़ྗ
    256x256x1
    Ϟσϧ
    Sigmoid
    
    

    View Slide

31. ReLU Sigmoid
    ׆ੑԽؔ਺
    
    

    View Slide

32. 256x256
    left : 0.6036
    top : 0.2854
    right : 0.8220
    bottom : 0.5236
    0.1092 : left
    0.4094 : top
    0.3209 : right
    0.6247 : bottom
    ࠲ඪͷ஋Ҭʢʣ
    
    

    View Slide

33. ReLU Sigmoid
    ׆ੑԽؔ਺
    
    

    View Slide

34. ܇࿅ɾֶश
    ϛχόον
    ֶश཰ʢ"EBN0QUJNJ[FSʣ
    
    

    View Slide

35. ݕग़݁Ռ
    
    

    View Slide

36. ̎ਓҎ্͍Δ৔߹͸ʁ
    
    

    View Slide

37. 
    

    View Slide

38. σʔλ಺༁
    
    0
    (18.57%)
    1
    (57.64%)
    2
    (11.93%)
    إͷ਺ ϑΝΠϧ਺ ׂ߹
    0 3748 18.57%
    1 11636 57.64%
    2 2408 11.93%
    3 844 4.18%
    4 596 2.95%
    5 325 1.61%
    6 214 1.06%
    7 108 0.53%
    

    View Slide

39. ೖྗͱग़ྗ
    256x256x1
    Ϟσϧ
    Sigmoid
    
    

    View Slide

40. إྖҬͷݕग़
    

    View Slide

41. إྖҬͷݕग़
    dconv
    3x3x1
    64
    dconv
    3x3x64
    64
    dconv
    3x3x256
    128 dconv
    3x3x64
    128
    dconv
    3x3x128
    128
    conv
    3x3x1
    64
    conv
    3x3x64
    64
    conv
    3x3x128
    256
    conv
    3x3x64
    128
    conv
    3x3x128
    128
    
    

    View Slide

42. إྖҬͷݕग़
    
    

    View Slide

43. ...
    ৞ΈࠐΈ૚ʢDPOWEʣͷ໾ׂ
    
    ಛ௃Ϛοϓ
    

    View Slide

44. إྖҬͷݕग़ʢ৞ΈࠐΈΦʔτΤϯίʔμʣ
    dconv
    3x3x3
    64
    dconv
    3x3x64
    64
    dconv
    3x3x256
    128 dconv
    3x3x64
    128
    dconv
    3x3x128
    128
    conv
    3x3x1
    64
    conv
    3x3x64
    64
    conv
    3x3x128
    256
    conv
    3x3x64
    128
    conv
    3x3x128
    128
    
    

    View Slide

45. ৞ΈࠐΈ૚ͷΈʹͨ͠৔߹
    conv
    3x3x1
    64
    conv
    3x3x64
    64
    conv
    3x3x128
    256
    conv
    3x3x64
    128
    conv
    3x3x128
    128
    dconv
    3x3x3
    64
    dconv
    3x3x64
    64
    dconv
    3x3x256
    128 dconv
    3x3x64
    128
    dconv
    3x3x128
    128
    
    

    View Slide

46. ৞ΈࠐΈ૚ͷΈʹͨ͠৔߹
    conv
    3x3x64
    stride 1
    conv
    3x3x64

    stride 1
    ReLU ReLU
    conv
    3x3x128

    stride 1
    conv
    3x3x128

    stride 1
    ReLU
    conv
    3x3x256

    stride 1
    conv
    3x3x256

    stride 1
    ReLU
    conv
    1x1x1

    stride 1
    256x256x1
    max_pool
    2x2
    stride 2
    max_pool
    2x2
    stride 2
    ReLU ReLU Sigmoid
    29x29x1
    
    

    View Slide

47. άϦουʢYʣ
    
    

    View Slide

48. άϦουͷͲ͜ʹإ͕͋Δ͔
    
    

    View Slide

49. ࠲ඪͷதԝͷηϧΛબ୒
    
    

    View Slide

50. ֤ηϧʹσʔλΛઃఆ
    
    confidence : 1.0
    center_x : 0.7128
    center_y : 0.4045
    width : 0.2184
    height : 0.2382
    1.0 : confidence
    0.2150 : center_x
    0.5169 : center_y
    0.2117 : width
    0.2151 : height
    confidence : 0.0
    center_x : 0.0
    center_y : 0.0
    width : 0.0
    height : 0.0
    

    View Slide

51. Ϩίʔυߏ଄
    ը૾σʔλʢόΠτྻʣ
    
    
    

    

    

    
    
    

    View Slide

52. Ϟσϧ
    conv
    3x3x64
    stride 1
    conv
    3x3x64

    stride 1
    ReLU ReLU
    conv
    3x3x128

    stride 1
    conv
    3x3x128

    stride 1
    ReLU
    conv
    3x3x256

    stride 1
    conv
    3x3x256

    stride 1
    ReLU
    conv
    1x1x5

    stride 1
    256x256x1
    max_pool
    2x2
    stride 2
    max_pool
    2x2
    stride 2
    ReLU ReLU Sigmoid
    29x29x5
    
    

    View Slide

53. ग़ྗ
    
    confidence
    center_x
    center_y
    width
    height
    29 × 29 × (4 + 1)
    = 4,205
    

    View Slide

54. ܇࿅ɾֶश
    ࠷దԽΞϧΰϦζϜ"EBN0QUJNJ[FS
    ֶश཰
    
    

    View Slide

55. Yͷݶք
    إதԝҎ֎ͷηϧͷDPOpEFODF΋ߴ͍
    DPOpEFODFͷߴ͍ηϧ͔Β

    ਖ਼͍͠࠲ඪ͕ಘΒΕͳ͍
    
    

    View Slide

56. ৞ΈࠐΈ૚ͷ໾ׂ
    conv
    3x3x64
    stride 1
    conv
    3x3x64

    stride 1
    ReLU ReLU
    conv
    3x3x128

    stride 1
    conv
    3x3x128

    stride 1
    ReLU
    conv
    3x3x256

    stride 1
    conv
    3x3x256

    stride 1
    ReLU
    conv
    1x1x1

    stride 1
    256x256x1
    max_pool
    2x2
    stride 2
    max_pool
    2x2
    stride 2
    ReLU ReLU Sigmoid
    29x29x1
    
    

    View Slide

57. ...
    લ૚͔ΒͷೖྗΛεΩϟϯͯ͠ಛ௃ϚοϓΛग़ྗ
    
    ಛ௃Ϛοϓ
    

    View Slide

58. ಛ௃Ϛοϓ͸ɺ
    খ͘͞ͳΔ΄ͲେҬͷಛ௃Λද͢
    
    

    View Slide

59. ಛ௃Ϛοϓ͸ɺ
    খ͘͞ͳΔ΄ͲେҬͷಛ௃Λද͢
    
    

    View Slide

60. దͨ͠αΠζͷηϧʹׂΓ౰ͯΔ
    
    

    View Slide

61. ৞ΈࠐΈΛ܁Γฦ͢աఔͰग़ྗ͢Δ
    
    Single Shot MultiBox Detector
    https://arxiv.org/abs/1512.02325
    

    View Slide

62. 
    Ϟσϧ
    

    View Slide

63. conv
    3x3x64
    stride 1
    conv
    3x3x64

    stride 1
    ReLU ReLU
    conv
    3x3x128

    stride 1
    conv
    3x3x128

    stride 1
    ReLU
    conv
    3x3x256

    stride 1
    conv
    3x3x256

    stride 1
    ReLU
    256x256x1
    max_pool
    2x2
    stride 2
    max_pool
    2x2
    stride 2
    ReLU ReLU
    29x29x256
    

    View Slide

64. conv
    3x3x256

    stride 2
    conv
    3x3x256

    stride 2
    conv
    3x3x256

    stride 2
    conv
    1x1x5

    stride 1
    Sigmoid
    conv
    1x1x5

    stride 1
    conv
    1x1x5

    stride 1
    Sigmoid
    conv
    1x1x5

    stride 1
    Sigmoid conv
    1x1x5

    stride 1
    Sigmoid
    1x1x5
    15x15x5
    8x8x5
    2x2x5
    4x4x5
    29x29x256
    conv
    1x1x5

    stride 1
    Sigmoid
    conv
    3x3x256

    stride 2
    conv
    3x3x256

    stride 2
    Sigmoid
    29x29x5
    

    View Slide

65. 
    ֶशσʔλ
    

    View Slide

66. ֶशσʔλ
    
    confidence : 1.0
    center_x : 0.7128
    center_y : 0.4045
    width : 0.2184
    height : 0.2382
    1.0 : confidence
    0.2150 : center_x
    0.5169 : center_y
    0.2117 : width
    0.2151 : height
    confidence : 0.0
    center_x : 0.0
    center_y : 0.0
    width : 0.0
    height : 0.0
    

    View Slide

67. ηϧͷׂΓ౰ͯج४
    ÷ = Jaccard
    Overlap
    ҰఆҎ্ͷ஋ʢॏͳΓʣͷηϧʹׂΓ౰ͯ
    ↓
    + - )
    (
    
    

    View Slide

68. EFG@MBOE SFDU
    
    XJEUISFDUSFDU
    IFJHIUSFDUSFDU
    JGXJEUIPSIFJHIU
    SFUVSO
    SFUVSOXJEUIIFJHIU
    EFGKBDDBSE@PWFSMBQ SFDU SFDU
    
    PWFSMBQ@MFGUNBY SFDU SFDU
    
    PWFSMBQ@SJHIUNJO SFDU SFDU
    
    PWFSMBQ@UPQNBY SFDU SFDU
    
    PWFSMBQ@CPUUPNNJO SFDU SFDU
    
    PWFSMBQ@MBOE

    View Slide

69. ޡࠩؔ਺
    ֬৴౓ʢ$POpEFODFʣˠ.4&ʢ.FBO4RVBSFE&SSPSʣ
    ࠲ඪʢ3FHJPOʣˠ4NPPUI--PTT
    ʢ'BTU3$//IUUQTBSYJWPSHBCTʣ
    
    

    View Slide

70. def _smooth_l1_loss(x):
    with tf.name_scope("smooth_l1"):
    abs_x = tf.abs(x)
    less_mask = tf.cast(abs_x < 1.0, tf.float32)
    return less_mask * (0.5 * tf.square(x)) + (1.0 - less_mask) * (abs_x - 0.5)
    def _loc_loss(logits, groundtruth):
    offset = logits - groundtruth
    return _smooth_l1_loss(offset)
    
    

    View Slide

71. MSEʢMean Squared Errorʣ Smooth L1 Loss
    
    

    View Slide

72. 
    ֶशʹࣦഊʜʜ
    

    View Slide

73. 1PTJUJWF4BNQMFͷׂ߹
    
    841
    29x29
    225
    15x15
    64
    8x8
    1
    4
    2x2
    16
    4x4
    1,151
    

    View Slide

74. 
    )BSE/FHBUJWF.JOJOH
    

    View Slide

75. 1PTJUJWFαϯϓϧʹՃ͑ͯɺޡࠩͷେ͖ͳαϯϓϧΛબ୒
    
    

    View Slide

76. /&("5*7&@$06/5@3"5&
    EFG@DBMD@MPTT@XJUI@IBSE@OFHBUJWF@NJOJOH MPTTFT QPTJUJWF@NBTL
    QPTJUJWF@DPVOU
    
    OFHBUJWF@DPVOUQPTJUJWF@DPVOU/&("5*7&@$06/5@3"5&
    CBUDI@TJ[FMPTTFTHFU@TIBQF
    >WBMVF
    QPTJUJWF@MPTTFTMPTTFTQPTJUJWF@NBTL
    OFHBUJWF@MPTTFTMPTTFTQPTJUJWF@MPTTFT
    UPQ@OFHBUJWF@MPTTFT @UGOOUPQ@L UGSFTIBQF OFHBUJWF@MPTTFT
    

    View Slide

77. 
    ֶशʹࣦഊʜʜ
    

    View Slide

78. إ͸্ख͘ݕग़Ͱ͖͍ͯΔ͕
    ࠲ඪ͕ਖ਼͘͠ͳ͍
    
    

    View Slide

79. 
    EFGBVMU#PY
    ʢ"ODIPSʣ
    

    View Slide

80. EFGBVMUCPYͷಋೖ
    
    left : 0.5000
    top : 0.2500
    right : 0.7500
    bottom : 0.5000
    0.0000 : left
    0.5000 : top
    0.2500 : right
    0.7500 : bottom
    

    View Slide

81. EFGBVMUCPYͱ࠲ඪ
    
    left : 0.5000
    top : 0.2500
    right : 0.7500
    bottom : 0.5000
    0.0000 : left
    0.5000 : top
    0.2500 : right
    0.7500 : bottom
    left : 0.6036
    top : 0.2854
    right : 0.8220
    bottom : 0.5236
    0.1092 : left
    0.4094 : top
    0.3209 : right
    0.6247 : bottom
    

    View Slide

82. EFGBVMUCPYͱ࠲ඪͷ
    ࠩʢP⒎TFUʣΛઃఆ
    
    left : 0.6036
    top : 0.2854
    right : 0.8220
    bottom : 0.5236
    left : 0.1036
    top : 0.0354
    right : 0.0720
    bottom : 0.0236
    0.1092 : left
    -0.0906 : top
    0.0709 : right
    -0.1253 : bottom
    0.1092 : left
    0.4094 : top
    0.3209 : right
    0.6247 : bottom
    

    View Slide

83. ग़ྗ ʹҙຯ͕ੜ·ΕΔ
    
    

    View Slide

84. ׆ੑԽؔ਺ͷมߋʢग़ྗ૚ʣ
    left : 0.1036
    top : 0.0354
    right : 0.0720
    bottom : 0.0236
    0.1092 : left
    -0.0906 : top
    0.0709 : right
    -0.1253 : bottom
    
    

    View Slide

85. ׆ੑԽؔ਺ͷมߋʢग़ྗ૚ʣ
    Sigmoid Hyperbolic Tangent
    
    

    View Slide

86. ֶशɾ܇࿅
    
    εςοϓʢόοναΠζ Ϛϧν(16ʣ

    ࠷దԽΞϧΰϦζϜ"EBN0QUJNJ[FS
    ֶश཰ ݮਰ TUFQ
    

    View Slide

87. ؀ڥʢ͘͞ΒͷߴՐྗίϯϐϡʔςΟϯάʣ
    $169FPO$PSFʷ
    .FNPSZ(#
    44%(#
    (F'PSDF(595*5"/9ʢ1BTDBMΞʔΩςΫνϟʣ(#ʷ
    (F'PSDF(595Jʢ1BTDBMΞʔΩςΫνϟʣ(#ʷ
    
    

    View Slide

88. ޡࠩͷมԽ
    
    

    View Slide

89. ݕূ
    
    

    View Slide

90. ݕূ
    
    

    View Slide

91. 
    ࠓޙͷ՝୊
    

    View Slide

92. ΞεϖΫτൺ͕େ͖͍ը૾΁ͷରԠ
    
    

    View Slide

93. ΞεϖΫτൺͷҟͳΔEFGBVMUCPYͷಋೖ
    
    

    View Slide

94. ٕज़ॻయ̏
    https://techbookfest.org/event/tbf03/
    circle/5686003050217472
    ೔࣌ɿ2017೥10݄22೔ʢ೔ʣ

    ɹɹɹ11࣌ʙ17࣌
    ৔ॴɿΞΩόɾεΫΤΞ
    ओ࠵ɿTechbooster / ୡਓग़൛ձ
    Illustration: ࠜઇΕ͍
    
    

    View Slide

95. ࢀߟ
    4JOHMF4IPU.VMUJ#PY%FUFDUPS

    IUUQTBSYJWPSHBCT
    44%4JOHMF4IPU.VMUJ#PY%FUFDUPSʢ೔ຊޠ༁ʣ2JJUB

    IUUQTRJJUBDPNEFUBJUFNTBFDFGDBB
    5FOTPS'MPXͰإݕग़ثΛࣗ࡞͢Δ͗͢ΌʔΜϝϞ

    IUUQNFNPTVHZBODPNFOUSZ
    5FOTPS'MPXͰ෺ମྖҬ༧ଌʢ3FHJPO1SPQPTBMʣΛࢼͯ͠ΈΔ

    IUUQXPSLQJMFTDPNUFOTPSqPXTTESFHJPOQSPQPTBM
    
    

    View Slide

96. ࢀߟ
    'BTU3$//

    IUUQTBSYJWPSHBCT
    'BTUFS3$//5PXBSET3FBM5*NF0CKFDU%FUFDUJPOXJUI3FHJPO1SPQPTBM/FUXPSLT

    IUUQTBSYJWPSHBCT
    :PV0OMZ-PPL0ODF6OJpFE 3FBM5JNF0CKFDU%FUFDUJPO

    IUUQTBSYJWPSHBCT
    +BDDBSE*OEFY

    IUUQTFOXJLJQFEJBPSHXJLJ+BDDBSE@JOEFY
    
    

    View Slide

97. C-LIS CO., LTD.
    ຊࢿྉ͸ɺ༗ݶձࣾγʔϦεͷஶ࡞෺Ͱ͢ɻܝࡌ͞Ε͍ͯΔΠϥετ͸ɺಛʹهࡌ͕ͳ͍৔߹͸ࠜઇΕ͍ͷஶ࡞෺Ͱ͢ɻ
    ຊࢿྉͷશ෦ɺ·ͨ͸Ұ෦ʹ͍ͭͯɺஶ࡞ऀ͔ΒจॻʹΑΔڐ୚Λಘͣʹෳ੡͢Δ͜ͱ͸ې͡ΒΕ͍ͯ·͢ɻ
    5IF"OESPJE4UVEJPJDPOJTSFQSPEVDFEPSNPEJpFEGSPNXPSLDSFBUFEBOETIBSFECZ(PPHMFBOEVTFEBDDPSEJOHUPUFSNTEFTDSJCFEJOUIF$SFBUJWF$PNNPOT"UUSJCVUJPO-JDFOTF
    ֤੡඼໊ɾϒϥϯυ໊ɺձ໊ࣾͳͲ͸ɺҰൠʹ֤ࣾͷ঎ඪ·ͨ͸ొ࿥঎ඪͰ͢ɻຊࢿྉதͰ͸ɺ˜ɺšɺäΛׂѪ͍ͯ͠·͢ɻ
    5IF"OESPJESPCPUJTSFQSPEVDFEPSNPEJpFEGSPNXPSLDSFBUFEBOETIBSFECZ(PPHMFBOEVTFEBDDPSEJOHUPUFSNTEFTDSJCFEJOUIF$SFBUJWF$PNNPOT"UUSJCVUJPO-JDFOTF
    https://speakerdeck.com/keiji/devfest-tokyo-2017-tensorflow
    

    View Slide