20171209 Sakura ML Night

Ece52fe9ce913851256726020707febd?s=47 Keiji ARIYAMA
December 09, 2017
Technology
0
120

20171209 Sakura ML Night

2017年12月9日に大阪で開催された「さくらの機械学習ナイト」の発表資料です。

「TensorFlowによるNSFW（職場で不適切な）画像検出」について。

Ece52fe9ce913851256726020707febd?s=128

Keiji ARIYAMA

December 09, 2017
Tweet

More Decks by Keiji ARIYAMA

See All by Keiji ARIYAMA
Ece52fe9ce913851256726020707febd?s=48 keiji
0
23
Ece52fe9ce913851256726020707febd?s=48 keiji
1
66
Ece52fe9ce913851256726020707febd?s=48 keiji
1
250
Ece52fe9ce913851256726020707febd?s=48 keiji
1
380
Ece52fe9ce913851256726020707febd?s=48 keiji
0
82
Ece52fe9ce913851256726020707febd?s=48 keiji
0
290
Ece52fe9ce913851256726020707febd?s=48 keiji
0
60
Ece52fe9ce913851256726020707febd?s=48 keiji
6
1.1k
Ece52fe9ce913851256726020707febd?s=48 keiji
1
240

Other Decks in Technology

See All in Technology
74948d1118cd84528f7861a3069dd8d9?s=48 qryuu
8
2.1k
B29f42636a5f1249b640473d49aa4514?s=48 linedevth
1
170
C786d7772602b488a7972c38c143f61c?s=48 con_mame
3
1.8k
Bf7fe621f4fe1615c228ef8a79b87282?s=48 shirayanagiryuji
2
730
0b238801605070def81a98cfe8061aee?s=48 shonansurvivors
1
180
53850955f15249a1a9dc49df6113e400?s=48 line_developers
1
290
7bcf6f29a0b9a348fe422a31a4f44752?s=48 cyberblack28
2
520
332f89cc697355902a817506b6995f2b?s=48 ytaka23
2
690
88f4e84b94fe07cddbd9e6479d689192?s=48 soudai
1
260
C744faee2ae82c1c4378e84d56daf5a8?s=48 yamamo
1
1.2k
404b6b310d5fc7680a62ef09a463a794?s=48 yutakakinjyo
0
220
1cf2de2da02f94ad87a7a31038721e6c?s=48 capsmalt
1
140

Featured

See All Featured
027d1ebf66cc039a0bd3b55eeadbe75d?s=48 sachag
448
35k
0c6fd6a08d0f898159cda7cafcacf07f?s=48 afnizarnur
175
14k
8081b26e05bb4354f7d65ffc34cbbd67?s=48 chriscoyier
775
240k
C157b16234f1e75e8eac3698c1d4414a?s=48 ddemaree
274
31k
027d1ebf66cc039a0bd3b55eeadbe75d?s=48 sachag
267
16k
E4f5d494ebdc9e7cce1aecf3ce3e8bc1?s=48 shpigford
164
18k
651dcfe41723207fbb0aa044a4f7c07f?s=48 dotmariusz
94
5k
A9704266587836f7e784235e5073b93e?s=48 jmmastey
2
230
5ce91fa49a613cbc3e20d5f96856473f?s=48 edds
55
9.2k
C620790ae5bf5b50c245b2e0ef95f338?s=48 deanohume
294
27k
766b9746b59e6f09e280cd33cf4ed419?s=48 skipperchong
4
480
9cde37f47e4a800ea081ea42de8d749a?s=48 dougneiner
55
5.3k

Transcript

  1. C-LIS CO., LTD.

  2. C-LIS CO., LTD.  ༗ࢁܓೋʢ,FJKJ"3*:"."ʣ $-*4$0 -5% "OESPJEΞϓϦ։ൃνϣοτσΩϧ Photo by

    Koji MORIGUCHI (MORIGCHOWDER) ػցֶश͸ͪΐͬͱ΍ͬͨ͜ͱ͋Γ·͢ Twitter΍ͬͯ·ͤΜ

  3. ͘͞ΒͷػցֶशφΠτ    5FOTPS'MPXͰ
 /4'8ը૾ݕग़

  4. 5FOTPS'MPXʢ೥݄ൃදʣ  ػց஌ೳ޲͚ܭࢉϑϨʔϜϫʔΫ ࠷৽όʔδϣϯʢ೥݄ʣ

  5.  ษڧձ΍Ζ͏ͥ

  6.  (PPHMF%FWFMPQFS(SPVQ

  7.  IUUQTHEHLPCFEPPSLFFQFSKQFWFOUT

  8.  Πϯλʔωοτ͔Β ޷Έͷը૾ΛࣗಈͰऩू͍ͨ͠

  9. © ࠜઇΕ͍ ؟ ڸ ͬ ່

  10. ؟ڸ່ͬ൑ఆ  1 0

  11. σʔληοτʢ೥݄࣌఺ʣ ؟ڸ່ͬɹຕ ඇ؟ڸ່ͬຕ  ؟ڸ່ͬ ඇ؟ڸ່ͬ ޡݕग़ ؟ڸ່ͬ ඇ؟ڸ່ͬ

  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 

  13. ߏ੒  Downloader  σʔληοτ Region + Label ઃఆ rsync

  14. ཧ૝ͷߏ੒  Downloader Face Detection Megane Detection ֬ೝɾमਖ਼ ೝࣝ݁Ռ ֶशʢ܇࿅ʣ

    λΠϜϥΠϯ ϝσΟΞ σʔληοτ ֶशʢ܇࿅ʣ TensorFlow  rsync

  15.  ௅ઓͷաఔΛಉਓࢽʹ

  16. ͞·͟·ͳ՝୊ σʔληοτ͕(#Λ௒͑ͨ͋ͨΓ͔ΒϩʔΧϧ΁ͷಉظ͕ࠔ೉ʹɻ ྖҬʢ3FHJPOʣͷઃఆͱϥϕϧͷ෇༩͸૝૾Ҏ্ʹෛՙ͕ߴ͍ɻ 

  17.  ը૾͕ສຕΛಥഁ σʔλ੔ཧ͕ࢸٸͷ՝୊ʹ

  18.  ໨ඪΛ࠶֬ೝ

  19.  Πϯλʔωοτ͔Β ޷Έͷ؟ڸ່ͬը૾ΛࣗಈͰऩू͍ͨ͠

  20. Ҏલͷߏ੒  Downloader  σʔληοτ Region + Label ઃఆ rsync

  21.  ྖҬʴϥϕϧ

  22. ৽͍͠ߏ੒  Downloader  σʔληοτ Tagઃఆ

  23.  λά megane girl

  24. ؟ڸ່ͬ൑ผϞσϧ  Ϟσϧ 1.00 0.00

  25.  %BUBTFU.BOBHFSGPS"OESPJE

  26.  σϞ

  27. https://twitter.com/35s_00/status/930366666973757441

  28.  https://twitter.com/_meganeco

  29.  /4'8ʢ/PU4BGF'PS8PSLʣ

  30. /4'8ը૾ 

  31. ͞·͟·ͳϦεΫ ࡞ۀͷϊΠζ ਫ਼ਆతͳෛՙ ๏తϦεΫ 

  32.  /4'8ը૾ͷݕग़

  33. ֶश༻σʔληοτʢ/4'8ʣ ਖ਼ྫɿ  ෛྫɿ   ← NSFWը૾

  34.  ܇࿅ɾֶश

  35. ڭࢣ༗Γֶश  ◦ × Ϟσϧ 1.00 0.00

  36. Ϟσϧͷߏ଄  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 output 1 256x256x1 max_pool 2x2 stride 2 max_pool 2x2 stride 2 ReLU ReLU Sigmoid max_pool 2x2 stride 2 conv 3x3x64
 stride 1 ReLU fc 768 ReLU bn bn bn

  37.  Sigmoid

  38. # モデル定義 NUM_CLASSES = 1 NAME = 'model3' IMAGE_SIZE =

    256 CHANNELS = 3 def prepare_layers(image, training=False): with tf.variable_scope('inference'): conv1 = tf.layers.conv2d(image, 64, [3, 3], [1, 1], padding='SAME', activation=tf.nn.relu, use_bias=False, trainable=training, name='conv1_1') conv1 = tf.layers.conv2d(conv1, 64, [3, 3], [1, 1], padding='VALID', activation=tf.nn.relu, use_bias=False, trainable=training, name='conv1_2') conv1 = tf.layers.batch_normalization(conv1, trainable=training, name='bn_1') 

  39. conv2 = tf.layers.conv2d(pool1, 128, [3, 3], [1, 1], padding='VALID', activation=tf.nn.relu,

    use_bias=False, trainable=training, name='conv2_1') conv2 = tf.layers.conv2d(conv2, 128, [3, 3], [1, 1], padding='VALID', activation=tf.nn.relu, use_bias=False, trainable=training, name='conv2_2') conv2 = tf.layers.batch_normalization(conv2, trainable=training, name='bn_2') pool2 = tf.layers.max_pooling2d(conv2, [2, 2], [2, 2]) 

  40. conv3 = tf.layers.conv2d(pool2, 256, [3, 3], [1, 1], padding='VALID', activation=tf.nn.relu,

    use_bias=False, trainable=training, name='conv4_1') conv3 = tf.layers.conv2d(conv3, 256, [3, 3], [1, 1], padding='VALID', activation=tf.nn.relu, use_bias=False, trainable=training, name='conv4_2') conv3 = tf.layers.batch_normalization(conv3, trainable=training, name='bn_4') pool3 = tf.layers.max_pooling2d(conv3, [2, 2], [2, 2]) conv = tf.layers.conv2d(pool3, 64, [1, 1], [1, 1], padding='VALID', activation=tf.nn.relu, use_bias=True, trainable=training, name='conv') return conv 

  41. def output_layers(prev, batch_size, keep_prob=0.8, training=False): flatten = tf.reshape(prev, [batch_size, -1])

    fc1 = tf.layers.dense(flatten, 768, trainable=training, activation=tf.nn.relu, name='fc1') fc1 = tf.layers.dropout(fc1, rate=keep_prob, training=training) output = tf.layers.dense(fc1, NUM_CLASSES, trainable=training, activation=None, name='output') return output 

  42. def _loss(logits, labels, batch_size, positive_ratio): cross_entropy = tf.nn.sigmoid_cross_entropy_with_logits( labels=labels, logits=logits)

    loss = tf.reduce_mean(cross_entropy) return loss def _init_optimizer(learning_rate): return tf.train.AdamOptimizer(learning_rate=learning_rate)  ޡࠩؔ਺ͱ࠷దԽΞϧΰϦζϜ

  43.  ֶशΛ্ख͘ਐΊΔ޻෉

  44. ਖ਼ྫɾෛྫͷൺ཰ ਖ਼ྫɿ  ෛྫɿ   ← NSFWը૾ NSFW

  45. def _hard_negative_mining(loss, labels, batch_size): positive_count = tf.reduce_sum(labels) positive_count = tf.reduce_max((positive_count,

    1)) negative_count = positive_count * HARD_SAMPLE_MINING_RATIO negative_count = tf.reduce_max((negative_count, 1)) negative_count = tf.reduce_min((negative_count, batch_size)) positive_losses = loss * labels negative_losses = loss - positive_losses top_negative_losses, _ = tf.nn.top_k(negative_losses, k=tf.cast(negative_count, tf.int32)) loss = (tf.reduce_sum(positive_losses / positive_count) + tf.reduce_sum(top_negative_losses / negative_count)) return loss  )BSE/FHBUJWF.JOJOH

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

  47. ֶश৚݅ ޡࠩؔ਺ަࠩΤϯτϩϐʔ ࠷దԽΞϧΰϦζϜ"EBN ֶश཰ όοναΠζ 

  48. 

  49. 

  50. طଘͷσʔληοτʹਪ࿦ʢJOGFSFODFʣΛ࣮ߦ  Downloader σʔληοτ Tagઃఆ inference trainer ֶशࡁΈϞσϧ ֶश༻σʔληοτ

  51. ਪ࿦݁Ռ /4'8  Ұൠը૾    NSFW
 8.6%

  52. ֶश༻σʔληοτʢ/4'8ʣ ਖ਼ྫɿ ɹˠɹ  ෛྫɿ ɹˠɹ  

  53. ܇࿅ɾֶशʹ͔͔Δܭࢉ࣌ؒ 

  54. 

  55.  σϞ (16ɾ$16ͷൺֱ

  56. $16ɾ(16ͷ଎౓ൺֱʢCBUDI4J[Fʣ 5*5"/9 TFDTUFQ 9FPO$PSF TFDTUFQ  ࠓճͷϞσϧͷֶशʹ͍ͭͯ͸ 5*5"/9ͷํ͕ഒ଎͍ʂ

  57. $16ɾ(16ͷ଎౓ൺֱʢCBUDI4J[F ʣ 5*5"/9  (595J  TFDTUFQ 9FPO$PSF TFDTUFQ 

    ࠓճͷϞσϧͷֶशʹ͍ͭͯ͸ (16ʷͷํ͕ഒ଎͍ʂ

  58.  ࠓޙͷ՝୊

  59. σʔληοταʔόʔͷ৴པੑ޲্ 

  60. JOGFSFODFʢਪ࿦ʣͷͨΊͷܭࢉࢿݯͷ֬อ  Downloader σʔληοτ Tagઃఆ inference trainer ֶशࡁΈϞσϧ ֶश༻σʔληοτ

  61. TAGS = [ 'original_art', 'nsfw', 'like', 'photo', 'illust', 'comic', 'face',

    'girl', 'megane',  ϥϕϧʢλάʣ 'school_uniform', 'blazer_uniform', 'sailor_uniform', 'gl', 'kemono', 'boy', 'bl', 'cat', 'dog', 'food', 'dislike', ]

  62. .PWJEJVT 

  63. ਪ࿦Λ.PWJEJVT΁Ҡߦ  Downloader σʔληοτ Tagઃఆ trainer ֶशࡁΈϞσϧ ֶश༻σʔληοτ inference

  64. Ϋϥε൑ఆϞσϧ  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 output 20 256x256x1 max_pool 2x2 stride 2 max_pool 2x2 stride 2 ReLU ReLU Sigmoid max_pool 2x2 stride 2 conv 3x3x64
 stride 1 ReLU fc 768 ReLU bn bn bn

  65. C-LIS CO., LTD. ຊࢿྉ͸ɺ༗ݶձࣾγʔϦεͷஶ࡞෺Ͱ͢ɻຊࢿྉͷશ෦ɺ·ͨ͸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/20171209-sakura-ml-night