ML Ops Study 2

C-LIS CO., LTD.

.-0QT4UVEZ BUΫοΫύουגࣜձࣾ 5FOTPS'MPXͷ܇࿅ࡁΈϞσϧΛ  "OESPJEΞϓϦʹࡌͤΔͱ͖ʹ  ۤ࿑ͨ͠࿩ 5FOTPS'MPXͰझຯͷը૾ऩूαʔόʔΛ࡞Δ೥݄߸

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

C-LIS CO., LTD.

C-LIS CO., LTD. ΍ͬ΂

ຊ൪؀ڥ΁ͷ౤ೖܦݧ͕ඞཁͳΒ ౤ೖͯ͠͠·͑͹͍͍͡Όͳ͍

؟ ڸ ͬ ່ ࠜઇΕ͍

޷Έͷ؟ڸ່ͬը૾Λ  ࣗಈͰऩू͍ͨ͠

̎Ϋϥε෼ྨ 1 0

ݱࡏͷγεςϜ

ධՁ༻ αʔόʔ ܇࿅ɾֶश༻αʔόʔ σʔληοτసૹ ʢTFRecordʣ ֶशࡁΈ ύϥϝʔλʔऔಘ ը૾औಘ
ը૾औಘ ϥϕϧ ෇͚ σʔληοτ؅ཧ αʔόʔ σʔλऩूݩ αʔϏε ը૾औಘ ϥϕϧ ෇͚ Android  ΞϓϦ

σʔληοτ؅ཧαʔόʔ $16"UISPO/FP()[ .FNPSZ(# 4UPSBHF44%(#  )%%5# 3"*%

σʔληοτ؅ཧαʔόʔͷ໾ׂ ը૾σʔλͷऩू ϝλσʔλʢΞϊςʔγϣϯɾϥϕϧʣͷ؅ཧ "1*ͷఏڙ ֶश༻σʔλʢ5'3FDPSEʣͷੜ੒

ը૾σʔλͷऩू

ϝλσʔλͷ؅ཧ

ϝλσʔλͷ؅ཧ label: 2 left: 283 top: 190 right:
435 bottom:301 = 1.0

ϥϕϧͷछྨ PSJHJOBM@BSU OTGX GBWPSJUF QIPUP JMMVTU DPNJD GBDF
GFNBMF NFHBOF TDISPPM@VOJGPSN CMB[FS@VOJGPSN TBJMPS@VOJGPSN HM LFNPOP NBMF CM DBU EPH GPPE EJTMJLF

"1*ͷఏڙ ը૾Ϧετͷऔಘ ը૾ͷऔಘ ը૾ͷݕࡧʢϥϕϧʣ ϥϕϧͷઃఆ ϥϕϧະઃఆը૾ͷϦετΛऔಘ σʔλऔಘݩ5XJUUFS*%ͷ௥Ճɾ࡟আ

ֶश༻σʔλͷੜ੒ 5'3FDPSEܗࣜPS+1&( $47ܗࣜ ը૾ͷϦαΠζ͸͜ͷஈ֊Ͱߦ͏ʢτϥϑΟοΫΛ௿ݮ $ python ./create_dataset.py \
--base_dir /dataset/source/ \ --output_dir ~/tfrecords_classifier \ --image_size 256 \ --tag_names megane,nsfw,favorite,illust

σʔλͷάϧʔϓԽ 0 1 2 3 4 5 6
7 8 9 σʔληοτ ςετσʔληοτ

ֶश༻αʔόʔ

{ "tag_name": "megane", "train_catalog_numbers": "0,1,2,3,4,5,6,7,8", "eval_catalog_numbers": "9", "data_augmentation": { "random_crop":
false, "random_colorize": true } } ֶश༻ͷઃఆϑΝΠϧ

ֶशͱݕূͷ࣮ߦ $ CUDA_VISIBLE_DEVICES=0,1 python ./train.py \ --learning_config config_megane.json
\ --tfrecords_dir ~/tfrecords_classifier \ --train_dir ~/train_single_discriminator \ --summary_dir ~/summary_single_discriminator \ --batch_size 64 \ --learning_rate 0.0001 \ --num_gpus 2 \ --max_step 100000 $ CUDA_VISIBLE_DEVICES=2 python ./eval.py \ --learning_config config_megane.json \ --tfrecords_dir ~/tfrecords_classifier \ --train_dir ~/train_single_discriminator \ --summary_dir ~/summary_single_discriminator $ tensorboard \ --logdir ~/summary_single_discriminator/megane/

ධՁ༻αʔόʔ

ධՁͷ࣮ߦ $ python3 client.py \ --tag_name megane \
--train_base_path ~/train_single_discriminator \ --train_file_name precision-0.956463/megane.ckpt-294000 \ --batch_size 16 \ --limit_batch 100

"OESPJEΞϓϦ

௚ۙͷ՝୊

ෆద੾ը૾͕ଟ͗͢Δ

ෆద੾ʢ/4'8ʣը૾ͱ؟ڸը૾ /4'8 positive: 36,083 → 7.17%  negative:
466,738 ؟ڸ positive: 23,559 → 2.44%  negative: 938,563

ෆద੾ը૾ϑΟϧλʔΛ ΞϓϦʹ૊ΈࠐΉ

NPEFM NSFW positive: 5,628 negative: 17,253

NPEFM ֶशࡁΈύϥϝʔλʔϑΝΠϧ ֶशࡁΈϞσϧ .pb 170MB

NPEFM@MJUF ֶशࡁΈύϥϝʔλʔϑΝΠϧ ֶशࡁΈϞσϧ .pb 10.7MB

NPEFM@MJUF

Ϟσϧͷߏ଄ .pb input result 128x128x3 1

private val IMAGE_WIDTH = 128 private val IMAGE_HEIGHT = 128
private val IMAGE_CHANNEL = 3 private val IMAGE_BYTES_LENGTH = IMAGE_WIDTH * IMAGE_HEIGHT * IMAGE_CHANNEL val imageByteBuffer: ByteBuffer = ByteBuffer.allocate(IMAGE_BYTES_LENGTH) val scaledBitmap = Bitmap.createScaledBitmap(bitmap, IMAGE_WIDTH, IMAGE_HEIGHT, false) scaledBitmap.copyPixelsToBuffer(imageByteBuffer) ը૾ΛόοϑΝʹ֨ೲ

val resultArray = FloatArray(1) fun recognize(imageByteArray: ByteArray): Float { val
start = Debug.threadCpuTimeNanos() tfInference.feed("input", imageByteArray, imageByteArray.size.toLong()) tfInference.run(arrayOf("result")) tfInference.fetch("result", resultArray) val elapsed = Debug.threadCpuTimeNanos() - start Log.d(TAG, "Elapsed: %d ns".format(elapsed)) return resultArray[0] } GFFESVOGFUDI

private val IMAGE_WIDTH = 128 private val IMAGE_HEIGHT = 128
private val IMAGE_CHANNEL = 3 private val IMAGE_BYTES_LENGTH = IMAGE_WIDTH * IMAGE_HEIGHT * IMAGE_CHANNEL val imageByteBuffer: ByteBuffer = ByteBuffer.allocate(IMAGE_BYTES_LENGTH) val scaledBitmap = Bitmap.createScaledBitmap(bitmap, IMAGE_WIDTH, IMAGE_HEIGHT, false) scaledBitmap.copyPixelsToBuffer(imageByteBuffer) scaledBitmap.recycle() ݪҼΒ͖͠΋ͷ ˢ"MQIBνϟϯωϧ͕ೖ͍ͬͯΔ

Ϟσϧͷߏ଄ .pb input result 128x128x4 1

with tf.Graph().as_default() as g: image_ph = tf.placeholder( tf.uint8, [model.IMAGE_SIZE *
model.IMAGE_SIZE * 4], name='input') image = tf.cast(image_ph, tf.float32) image = tf.reshape( image, [model.IMAGE_SIZE, model.IMAGE_SIZE, 4]) image = image[:, :, :3] QCग़ྗ࣌ʹDIΛड͚ೖΕΔΑ͏ʹάϥϑΛมߋ

'PPE(BMMFSZ https://github.com/keiji/food_gallery_with_tensorﬂow ΪϟϥϦʔʹอଘ͞Ε͍ͯΔ৯෺ͷը૾Λදࣔ http://techlife.cookpad.com/entry/2017/09/14/161756 ΫοΫύου։ൃऀϒϩά ྉཧ͖Ζ͘ʹ͓͚Δྉཧʗඇྉཧ൑ผϞσϧͷৄࡉ

૊ΈࠐΜͰΈ͚ͨΕͲɺ ਫ਼౓͸͋·Γߴ͘ͳ͍ʜʜ

ࠓޙͷ՝୊ 5FOTPS'MPX-JUF΁ͷҠߦ ߴਫ਼౓ͷϞσϧͷѹॖʢল༰ྔԽʣ 1SVOJOH 2VBOUJ[BUJPO %JTUJMMBUJPO ML Kit:
Machine Learning SDK for mobile developers (Google I/O '18) https://youtu.be/Z-dqGRSsaBs?t=32m10s

ࠓޙͷ՝୊ ֶशαΠΫϧΛࣗಈԽ͍͖͍ͯͨ͠ɻ ʢఆظతʹֶशσʔλͷੜ੒ͱసૹΛߦ͍ɺ࠶ֶश͢ΔͳͲʣ σʔλϕʔεͷߴ଎Խ  ɹϥϕϧݕࡧ͕ඇৗʹ஗͍ͷ͕՝୊ɻઃܭΛݟ௚͢ඞཁ͋Γ  ɹݕࡧΠϯσοΫεͷ(PPHMF$MPVE4UPSFҠߦΛݕ౼ 

ࠓޙͷ՝୊ σʔληοτ؅ཧαʔόʔͷߋ৽ 5FOTPS'MPXΑΓɺ$16ʹ*OUFM"79͕ඞਢʹͳͬͨʢQJQ൛ʣ

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͸ɺɺɺäΛׂѪ͍ͯ͠·͢ɻ 5IF"OESPJESPCPUJTSFQSPEVDFEPSNPEJpFEGSPNXPSLDSFBUFEBOETIBSFECZ(PPHMFBOEVTFEBDDPSEJOHUPUFSNTEFTDSJCFEJOUIF$SFBUJWF$PNNPOT"UUSJCVUJPO-JDFOTF

ML Ops Study 2

ML Ops Study 2

More Decks by Keiji ARIYAMA

Other Decks in Technology

Featured

Transcript