【再々開催】Raspberry_PiとUSBカメラで_はじめようディープラーニング20191207

ももち浜TECHカフェ
「【再々開催】Raspberry_PiとUSBカメラで_はじめようディープラーニング20191207」

ももち浜TECHカフェは福岡市ももち浜から発信する、IoT・AR・VRなどをテーマにしたハンズオン企画です。
今回のテーマはRaspberry Pi ＆ 機械学習・ディープラーニング。追加開催版です。

Raspberry Piはイギリスのラズベリーパイ財団が開発しているシングルボードタイプのコンピュータです。安価で小型なためIoT機器への組込み・プロトタイピング等で日本でも趣味・業務問わず広く普及が進んできています。

今回は、そんなRaspberry PiとUSBカメラを使って、機械学習やディープラーニングの初歩を学んでいただけます。

Transcript

1. 3BTQCFSSZ
   1Jͱ
   64#ΧϝϥͰ͸͡ΊΑ͏
   σΟʔϓϥʔχϯά 
2. ࣗݾ঺հ .BLF-4* 73"3 ࢁాඒึʢ΍·ͩΈ΄ʣ
   ɾʮ*P5Λ΋ͬͱ਎ۙʹɺ΋ͬͱָ͘͠ɻະདྷ΁ʯ
   ɹɹͰ͓ͳ͡Έͷɺגࣜձࣾ$FOUFS2
   ୅දऔక໾
   ɾʮ*P5ओ්ʯͱ໊৐͍ͬͯ·͢
   ɾঁࢠͩΒ͚ͷిࢠ޻࡞
   ओ࠵
   ɾझຯɿϋϯζΦϯ΍ษڧձʹࢀՃ͢Δ͜ͱɻ
   ɹ
   
   
3. ΈΜͳͷࣗݾ঺հλΠϜ ̍ʣ໊͓લ ̎ʣԿΛ͍ͯ͠Δਓ͔ ̏ʣࠓ೔ͷϞνϕʔγϣϯ
4. ຊ೔ֶΜͰ͍͚ͨͩΔ͜ͱ ̍ʣRaspberryPiͷجૅ஌ࣝ ̎ʣσΟʔϓϥʔχϯάʹඞཁͳϥΠϒϥϦʔ౳ͷΠϯετʔϧखॱ ̏ʣσʔληοτMNISTΛ࢖ͬͨखॻ͖ೝࣝจࣈ ̐ʣRaspberryPiͰUSBΧϝϥΛ࢖͏खॱ ̑ʣֶशࡁΈϞσϧInceptionV3Λ࢖ͬͨը૾ೝࣝ ̒ʣը૾Λ࢖ͬͯσΟʔϓϥʔχϯά(CNN)ͤ͞Δϓϩηε
5. 3BTQCFSSZ
   1Jͱ
   64#ΧϝϥͰ͸͡ΊΑ͏
   σΟʔϓϥʔχϯά
6. 3BTQCFSSZ1JͷجૅΛֶͿ 3BTQCFSSZ
   1Jͷछྨ
7. 3BTQCFSSZ1JͷجૅΛֶͿ 3BTQCFSSZ
   1Jͷछྨ ͋ͨΒ͘͠
   ൃച͞Ε·ͨ͠ʂ
   ൢചՁ֨͸
   ԁʙ
8. ٕదະऔಘػثΛ༻͍࣮ͨݧ౳ͷ
   ಛྫ੍౓ͷਃ੥͖ͯ͠·ͨ͠ʂʂ
9. ຊ೔ͷ࢖༻ػث 3BTQCFSSZ1JͷجૅΛֶͿ 3BTQCFSSZ
   1Jͷछྨ
10. ຊ೔࢖༻͢Δ3BTQCFSSZ1J
    
    .PEFM#͸
    ɹ"3.
    $PSUFY"ʢίΞɺ
    ()[ʣ
    ɹ #SPBEDPN
    #$.
    
    
    
    ετϨʔδɹNJDSP4%
    ɹ64#
    Y
    
    
    
    )%.*
    º
    
    
    
    -"/
    #"4&5
    ɹΦʔσΟΦɹNNεςϨΦ
    ɹిݯɹ7ʗ"ʢ64#
    NJDSP#ʣ
    ɹେ͖͞ɹºNN
    ɹ8J'J͸()[ଳͷΈ࢖༻Ͱ͖·͢ɻ
    3BTQCFSSZ1JͷجૅΛֶͿ
    3BTQCFSSZ
    1J
    ̏
    .PEFM#ͱ͸ ˞3BTQCFSSZ1J͸օ͞Μ͕࢖͍ͬͯΔ1$ΑΓ൓Ԡ଎౓͕ͱͯ΋஗͍Ͱ͢ɻ
    ɹʢ͙͢൓Ԡ͠ͳ͍͔ΒͱԿ౓΋ΫϦοΫͨ͠ΓίϚϯυΛ࠶౓ೖྗ͠ͳ͍ʣ
11. https://www.raspberrypi.org/ 3BTQCFSSZ1Jެࣜϖʔδ ֤छΠϯετʔϧ΍࢖͍ํͳͲϥζύΠͷ࠷৽৘ใ͸ͪ͜ΒΛνΣοΫ͠·͢ɻ 3BTQCFSSZ1JͷجૅΛֶͿ
12. https://www.raspberrypi.org/downloads/ ࠓճͷηϛφʔͰ͸
    طʹ04͕Πϯετʔϧ
    ͞ΕͨNJDSP4% (# Λ
    ࢖͍·͢ɻ
    ͳͷͰΠϯετʔϧͷ
    ࡞ۀ͸ෆཁͰ͢ɻ
    ϥζύΠ͸3BTQCJBOʢϥζϏΞϯʣ
    ͱ͍͏-JOVYܥͷ04Λ࢖ͬͯಈ͔͠·͢ɻ
    /00#4 ψʔϒεʣͱ͍͏04ͷΠϯετʔϥʔΛ࢖͏ͱศརͰ͢ɻ 3BTQCFSSZ
    1Jͷ04ʹ͍ͭͯʢNJDSP4% 3BTQCFSSZ1JͷجૅΛֶͿ
13. 3BTQCFSSZ1JͱσΟʔϓϥʔχϯάͰͲΜͳ͜ͱ͕Ͱ͖Δͷʁɹɹ ͖Ύ͏Γͷ඼࣭൑ผ ͔Β͋͛ͷ഑હ
14. ख़࿅ͨ͠ਓͷϊ΢ϋ΢ˠࣗಈԽ ˁ ͦΕͰ͸ɺ͓଴ͨͤ͠·ͨ͠ʂ
    ϥζύΠΛҰॹʹ४උ͍͖ͯ͠·͠ΐ͏ʂ Έͳ͞Μ͸ɺԿΛͭ͘Γ·͔͢ʁ
15. ຊ೔ͷ3BTQCFSSZ1Jपลػثߏ੒ͱ઀ଓ֬ೝ 3BTQCFSSZ1J
    ̏
    .PEFM
    # 64#ΩʔϘʔυ 64#Ϛ΢ε 64#Χϝϥ )%.*
    σΟεϓϨΠ ిݯ͸·ͩೖΕͳ͍Ͱ σΟεϓϨΠ
    ɹిݯ NJDSP4% -"/έʔϒϧ
16. 3BTQCFSSZ1JͷిݯΛೖΕͯىಈ ˞ͻͱੲલ3BTQCFSSZ1J͕ྲྀߦͬͨ࣌ͱ͸ҧͬͯࠓ͸ࣗಈͰϩάΠϯ͠·͢
17. ˞
    
    
    
    
    
    
    
    
    
    
    ͷγεςϜͷจࣈ͕খͯ͘͞ݟʹ͍͘ͳ͋ʝͱ͍͏ਓ͸ઃఆมߋ 3BTQCFSSZ1Jͷจࣈʹ͍ͭͯ എܠը໘ΛӈΫϦοΫ
18. γεςϜͷจࣈ͕খ͍࣌͞͸എܠը໘ΛӈΫϦοΫ ʮσεΫτοϓͷઃఆʯΛબ୒ͨ͠Β"QQFBSBODF
    4FUUJOH͕ ։͘ͷͰɺʮ4ZTUFNʯͷ'POUͷϘλϯΛΫϦοΫ͢Δͱ
    ʮ1JDL
    B
    'POUʯ͕։͘ͷͰݟ΍͍͢4J[Fʹมߋ͍ͩ͘͞ɻ
19. ɹˢίίΛΫϦοΫ͢Δͱ
    ɹɹ-95FSNJOBM্ཱ͕͕ͪΔ
    -95FSNJOBMͷىಈ
20. λʔϛφϧͷจࣈ͕খ͍࣌͞͸ฤूΛΫϦοΫ ʮઃఆ 4 ʯΛબ୒ͨ͠Β৽͘͠΢Οϯυ΢͕։͘ͷͰɺ
    ʮελΠϧʯͷʰ୺຤ͷϑΥϯτʱͷϘλϯ෦෼ΛΫϦοΫ͢Δͱ
    ʮϑΥϯτͷબ୒ʯ͕։͘ͷͰݟ΍͍͢4J[Fʹมߋ͍ͩ͘͞ɻ
21. ࠓճ࢖༻͢ΔϓϩάϥϜݴޠ͸ʝ ✋QZUIPOΛ৮ͬͨ͜ͱ͕͋Δํ͸खΛ্͛ͯ ίʔυ͕γϯϓϧͰɺ૊ΈࠐΈ΍8FCΞϓϦͷ։ൃɺ·ͨਓ޻஌ೳ΍ɺ
    ϏοάσʔλղੳͳͲͰΑ͘࢖༻͞ΕΔ൚༻ϓϩάϥϛϯάݴޠͰ͢ɻ ʮ1ZUIPOʢύΠιϯʣͰ͢ʯ
22. ίϚϯυΛೖྗͯ͠Έ·͠ΐ͏ʂ 
    ίϚϯυ໊ ˢˈϚʔΫ͸දࣔ͞Ε͍ͯΔͷͰɺˈ͸ೖΕΔඞཁ͕ͳ͍Ͱ͢ɻ -95FSNJOBMͷը໘ʹίϚϯυΛೖྗ͢Δ
23. 5FOTPS'MPXͱ͸ 5FOTPS'MPXʢςϯιϧϑϩʔʣͱ͸ɺ͞·͟·ͳػցֶश ʢ.BDIJOF
    -FBSOJOHʣͷ෼໺Ͱ࢖༻͞Ε͍ͯΔ044ʢΦʔϓ ϯιϑτ΢ΣΞϥΠϒϥϦʣͰ͢ɻ(PPHMF͕:PV5VCF΍ (PPHMF຋༁ɺԻ੠ݕࡧͳͲͰ࢖༻͍ͯͨ͠πʔϧͰɺ 3BTQCFSSZ1J্Ͱ΋ಈ࡞͠·͢ɻ ,FSBTͱ͸ ,FSBTʢέϥεʣͱ͸ɺ1ZUIPOͰॻ͔Εͨχϡʔϥϧωοτ ϫʔΫͷϥΠϒϥϦͰ͢ɻ5FOTPS'MPXɺ$/5,ʢ.JDSPTPGU
    $PHOJUJWF
    5PPMLJUʣɺ5IFBOPʢςΞϊʣ্Ͱಈ࡞͢Δ͜ͱ͕ Ͱ͖·͢ɻࠓճ͸,FSBTΛ࢖ͬͯը૾ೝࣝΛ͓͜ͳ͍·͢ɻ IUUQTLFSBTJP ࠓճ࢖༻͢Δ̎ͭͷπʔϧ͸ʝ
24. σΟʔϓϥʔχϯάʹඞཁͳ
    ϥΠϒϥϦ౳ͷΠϯετʔϧखॱ طʹΠϯετʔϧࡁΈ
    ͳͷͰɺྲྀΕ͚ͩΛ
    આ໌͠·͢ɻ
25. ݱࡏͷ1ZUIPOͱQJQͷόʔδϣϯ֬ೝ ̍ʣ1ZUIPOʢ̏ܥʣͷόʔδϣϯΛ֬ೝ͢ΔίϚϯυ ̎ʣQJQ QZUIPOͷόʔδϣϯ؅ཧπʔϧʣ̏ܥͷόʔδϣϯΛ֬ೝ͢ΔίϚϯυ QJQ
    
    GSPN
    VTSMJCQZUIPOEJTUQBDLBHFTQJQ
    (QZUIPO
    ) 
    QZUIPO
    
    7 
    QJQ
    
    7 1ZUIPO
     1ZUIPOͷόʔδϣϯ͸̎ܥͱ ̏ܥ͕͋Γ·͢ɻࠓճ͸ܥΛ ѻ͍·͢ɻ7͚ͩେจࣈͰ͢ʂ QJQ͸ύοέʔδ؅ཧγεςϜͰ
    ͢ɻQJQ͸1ZUIPȌܥͷ؅ཧ
26. "5-"4ʢઢܗ୅਺ϥΠϒϥϦʣͷΠϯετʔϧ
    ɹɹ "VUPNBUJDBMMZ
    5VOFE
    -JOFBS
    "MHFCSB
    4PGUXBSF
    ʣ 
    TVEP
    BQU
    JOTUBMM
    MJCBUMBTCBTFEFW "5-"4ʢઢܗ୅਺ϥΠϒϥϦʣͷΠϯετʔϧखॱ
    ్தͰΠϯετʔϧ͕ࢭ·ͬͯ
    ଓߦ͠·͔͢ʁ<:O>ͱฉ͔Ε·͢ͷͰ
    :Λೖྗޙ&OUFSͯ͠
    ଓ͚͍ͯͩ͘͞ɻ
    ஫ҙɿ
    طʹΠϯετʔϧࡁΈͰ͢
    ίϚϯυΛೖྗ͢Δඞཁ͸͋ Γ·ͤΜɻ
27. ȊQZͷΠϯετʔϧखॱ IQZ ΤΠνϑΝΠϒύΠ ͷΠϯετʔϧ QZUIPOͰ )%' )JFSBSDIJDBM
    %BUB
    'PSNBU ܗࣜΛ ѻ͑ΔΑ͏ʹ͠·͢ɻ 
    TVEP
    BQUHFU
    JOTUBMM
    QZUIPOIQZ ్தͰΠϯετʔϧ͕ࢭ·ͬͯ
    ଓߦ͠·͔͢ʁ<:O>
    ͱฉ͔Ε·͢ͷͰ
    :Λೖྗޙ&OUFSͯ͠
    ଓ͚͍ͯͩ͘͞ɻ
    ˞LFSBTͳͲͷχϡʔϥϧωοτϫʔΫϥΠϒϥϦͷֶश݁ՌΛอଘ͢ΔϑΥʔϚοτ ஫ҙɿ
    طʹΠϯετʔϧࡁΈͰ͢
    ίϚϯυΛೖྗ͢Δඞཁ͸͋ Γ·ͤΜɻ
28. 5FOTPSqPXͷΠϯετʔϧखॱ 5FOTPSqPXͷΠϯετʔϧ 
    TVEP
    QJQ
    JOTUBMM
    UFOTPSqPX ʢ͜͜·Ͱ෼ఔ౓ʣ
    3VOOJOH
    TFUVQQZ
    CEJTU@XIFFM
    GPS
    OVNQZ
    ʜ
    
    ʢ͜ͷ͋ͱ͞Βʹ̎̌෼ఔ౓͔͔Γ·͢ɻʣ
    ஫ҙɿ
    طʹΠϯετʔϧࡁΈͰ͢
    ίϚϯυΛೖྗ͢Δඞཁ͸͋ Γ·ͤΜɻ
29. ,FSBTͷΠϯετʔϧखॱ ,FSBTͷΠϯετʔϧ 
    TVEP
    QJQ
    JOTUBMM
    LFSBT ஫ҙɿ
    طʹΠϯετʔϧࡁΈͰ͢
    ίϚϯυΛೖྗ͢Δඞཁ͸͋ Γ·ͤΜɻ
30. 0QFO$7ͷΠϯετʔϧखॱ 0QFO$7ͱ͸ 
    TVEP
    BQUHFU
    JOTUBMM
    QZUIPOPQFODW ஫ҙɿ
    طʹΠϯετʔϧࡁΈͰ͢
    ίϚϯυΛೖྗ͢Δඞཁ͸͋ Γ·ͤΜɻ ͜ΕͰඞཁͳΠϯετʔϧ͕׬ྃ͠·ͨ͠ʂ ը૾΍ಈըΛॲཧ͢Δػೳ͕࣮૷͞Ε͍ͯΔϥΠϒϥϦ
31. ͜ΕͰUFOTPSqPXͱ,FSBTΛ࢖͏ ؀ڥ͕੔͍·ͨ͠ʂʂ ͜ΕͰ3BTQCFSSZ1JͰσΟʔϓϥʔχϯά͕ࢼͤ·͢ʂ
    
32. ਓ޻஌ೳͱػցֶशͱਂ૚ֶशͷؔ܎ ਓ޻஌ೳ ػցֶश ਂ૚
    ֶश ೖྗ ग़ྗ ʢػցʹίϯϐϡʔλʹֶशͤ͞Δ͜ͱʣ ʢόΠΦςΫϊϩδʔ
    ɹɹͳͲ΋͋Δʣ
33. σΟʔϓϥʔχϯάʢਂ૚ֶशʣͱ͸ σΟʔϓϥʔχϯάͱ͸Ի੠ͷೝࣝɺը૾ೝࣝɺҟৗݕ஌ͳͲ͢ΔͨΊʹɺίϯϐϡʔ λʔʹֶशͤ͞Δͻͱͭͷख๏Ͱ͢ɻਓؒͷ೴ͷதʹ͋ΔχϡʔϩϯΛਅࣅͨχϡʔϥϧ ωοτϫʔΫͱݺ͹ΕΔख๏ͷ֊૚ΛਂΊͨΞϧΰϦζϜʹͳΓ·͢ɻ தؒ૚ͷ֊૚͕ΑΓਂ͘ͳ͍ͬͯΔ ਓؒͷ೴ͷதʹ͋Δχϡʔϩϯ ܗࣜχϡʔϩϯ В y x1 x2 x3 xN w1 w2 w3 wN …………… …… σΟʔϓϥʔχϯά
    ೖྗ૚
    
    
    
    
    
    
    
    
    
    
    
    தؒ૚ɹ
    
    
    
    
    
    
    
    
    
    
    ग़ྗ૚ χϡʔϥϧωοτϫʔΫ
    ೖྗ૚
    
    
    தؒ૚ɹग़ྗ૚
34. ը૾ೝࣝͤ͞Δҝͷ̎ͭͷϑΣʔζ ࣗಈͰը૾ೝࣝΛͤ͞Δʹ͸େ͖͘ೋͭͷϑΣʔζ͕͋Γ·͢ɻ
    ୈ̍ϑΣʔζɿֶशϞσϧΛͭ͘Δ ୈ̎ϑΣʔζɿֶशࡁΈϞσϧΛ࢖͏ σʔλ
    ॲཧ σʔλ
    ऩू ػցֶश
    ਂ૚ֶश ֶशͤ͞Δը૾ ֶशࡁΈ
    Ϟσϧ ֶश༻
    σʔληοτ
    ೝ͍ࣝͤͨ͞ը૾ ೖྗ ग़ྗ ֶशࡁΈ
    Ϟσϧ ਪ࿦ ը૾ೝࣝΛ
    ࢖ͬͨॲཧ
35. ·ͣ͸ֶशϞσϧΛͭ͘ΔϑΣʔζΛֶ΅͏ʂ ࣗಈͰը૾ೝࣝΛͤ͞Δʹ͸େ͖͘ೋͭͷϑΣʔζ͕͋Γ·͢ɻ
    ୈ̍ϑΣʔζɿֶशϞσϧΛͭ͘Δ ୈ̎ϑΣʔζɿֶशࡁΈϞσϧΛ࢖͏ σʔλ
    ॲཧ σʔλ
    ऩू ػցֶश
    ਂ૚ֶश ֶशͤ͞Δը૾ ֶशࡁΈ
    Ϟσϧ ֶश༻
    σʔληοτ
    ೝ͍ࣝͤͨ͞ը૾ ೖྗ ग़ྗ ֶशࡁΈ
    Ϟσϧ ਪ࿦ ը૾ೝࣝΛ
    ࢖ͬͨॲཧ
36. ·ͣ͸ֶशϞσϧΛͭ͘ΔϑΣʔζΛֶ΅͏ʂ ਓखෆ଍Ͱ
    ༣ศ෺Λ࢓෼͚Δ ͷ͕
    େมͰ͢ɻ
    ͋ͳͨͳΒɺͲ͏΍ͬͯ༣ศہͷനϠΪ͞ΜΛॿ͚·͔͢ʁʁ ༣ศ൪߸Ͱ
    ࢓෼͚͍ͯ·͢ɻ
    
37. ·ͣ͸ֶशϞσϧΛͭ͘ΔϑΣʔζΛֶ΅͏ʂ खॻ͖༣ศ൪߸ͷ
    ਺ࣈΛࣗಈͰೝࣝ
    ͤ͞Δͱָͩͳʂ
    ·ͣ͸͍ΖΜͳਓͷ
    खॻ͖਺ࣈΛ
    ूΊΑ͏ʂ ͏Θʙ࢓෼͚Δ ྔ͕ଟͯ͘େม ͩ͠ɺखॻ͖਺ ࣈͬͯಡΈͮΒ ͍ͳ͋ɻ खॻ͖਺ࣈʢ༣ศ൪߸ʣΛը૾ೝࣝͰ͖ΔΑ͏ʹ͠Α͏ʂ
38. ·ͣ͸ֶशϞσϧΛͭ͘ΔϑΣʔζΛֶ΅͏ʂ ֶशϞσϧ σʔληοτ
    ɹ࡞੒ ֶश༻σʔλɹɹɹɹɹɹςετ༻σʔλ ूΊ͖ͯͨखॻ͖਺ࣈσʔλ ूΊͨσʔλ͔ΒσʔληοτΛ࡞੒ֶͯ͠शϞσϧΛ࡞Δ खॻ͖਺ࣈʢ༣ศ൪߸ʣΛը૾ೝࣝͰ͖ΔΑ͏ʹ͠Α͏ʂ ֶश σʔλॲཧ
39. ·ͣ͸ֶशϞσϧΛͭ͘ΔϑΣʔζΛֶ΅͏ʂ ֶश༻σʔλɹɹɹɹɹɹςετ༻σʔλ ूΊ͖ͯͨखॻ͖਺ࣈσʔλ ֶशϞσϧ ςετ༻σʔλͰ
    ਖ਼౴཰Λௐ΂Δ ϓϩάϥϛϯά ςετ ೲ඼ ׬
    ੒ ૊ΈࠐΈ
40. खॻ͖਺ࣈΛը૾ೝࣝͤ͞ΔͨΊͷ޻ఔ ୈ̍ϑΣʔζɿֶशϞσϧΛͭ͘Δ σʔλ
    ॲཧ σʔλ
    ऩू ػցֶश
    ਂ૚ֶश ֶशͤ͞Δը૾ ֶशࡁΈ
    Ϟσϧ ֶश༻
    σʔληοτ
    ֶशϞσϧ
41. खॻ͖਺ࣈͷσʔληοτ./*45ʢΤϜχετʣ ػցֶशΛֶͿਓʹͱͬͯ௒༗໊ͳσʔληοτͰ͢ɻ͔Β· Ͱͷສݸͷखॻ͖਺ࣈʢֶश༻̒ສݸɾςετ༻̍ສݸʣͷը૾ ͱͦΕʹରԠ͢Δ౴͑ʢϥϕϧʣ͕ఏڙ͞Ε͍ͯ·͢ɻ
42. ./*45ͷը૾ɹɹɹɹը૾Λd·Ͱͷ਺ࣈͰදݱͨ͠ߦྻ
43. 5IPOOZ 1ZUIPO*%&1ZUIPOͷॳ৺ऀ༻౷߹؀ڥ Λ࢖͏ UIPOOZͱ͍͏σϑΥϧτͰ
    ೖ͍ͬͯΔQZUIPOͷ*%&Λ
    ࠓճ࢖͍·͢ɻ
    ./*45ΛಡΈࠐΉϓϩάϥϜ
    NOJTUQZΛ
    ࣄલʹೖΕ͓͖ͯ·ͨ͠ͷͰ
    ։͖·͢ɻ
    
    UIPOOZ
    NOJTUQZ
    ͱλʔϛφϧͰଧ͍ͬͯͩ͘͞ɻ
    खॻ͖਺ࣈͷσʔληοτ./*45
44. from keras.datasets import mnist import matplotlib.pyplot as plt (X_train, y_train), (X_test, y_test) = mnist.load_data() print("label",y_train[0]) plt.imshow(X_train[0].reshape(28,28),cmap='Greys') plt.show() खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖͓͏
45. from keras.datasets import mnist import matplotlib.pyplot as plt (X_train, y_train), (X_test, y_test) = mnist.load_data() print("label",y_train[0]) plt.imshow(X_train[0].reshape(28,28),cmap='Greys') plt.show() खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖͓͏ LFSBTʹ༻ҙ͞Εͯ ͍Δ./*45ΛΠϯ ϙʔτͯ͠ɺը૾Λ දࣔͰ͖ΔΑ͏ʹ QMUͱઃఆ͢Δ
46. from keras.datasets import mnist import matplotlib.pyplot as plt (X_train, y_train), (X_test, y_test) = mnist.load_data() print("label",y_train[0]) plt.imshow(X_train[0].reshape(28,28),cmap='Greys') plt.show() खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖͓͏ NOJTUΛಡΈࠐΉ
47. from keras.datasets import mnist import matplotlib.pyplot as plt (X_train, y_train), (X_test, y_test) = mnist.load_data() print("label",y_train[0]) plt.imshow(X_train[0].reshape(28,28),cmap='Greys') plt.show() खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖͓͏ NOJTUͷ࠷ॳͷσʔ λͷϥϕϧͱը૾ Λग़ྗ
48. 5IPOOZ 1ZUIPO*%&1ZUIPOͷॳ৺ऀ༻౷߹؀ڥ Λ࢖͏ 3VOϘλϯΛԡ࣮ͯ͠ߦͤ͞Α͏ʂ खॻ͖਺ࣈͷσʔληοτ./*45 &Run mnist1.py Using TensorFlow backend. WARNING: Logging before flag parsing goes to stderr. W0823 09:50:29.745067 1995954896 deprecation_wrapper.py:118] From /home/pi/.local/lib/python3.7/site- packages/tensorflow/__init__.py:98: The name tf.AUTO_REUSE is deprecated. Please use tf.compat.v1.AUTO_REUSE instead. WARNING͕͍͔ͭ͘ग़·͕͢ ໰୊ͳ͍ͷͰ͠͹Β͘଴ͪ·͢
49. from keras.datasets import mnist import matplotlib.pyplot as plt (X_train, y_train), (X_test, y_test) = mnist.load_data() print("label",y_train[0]) plt.imshow(X_train[0].reshape(28,28),cmap='Greys') plt.show() खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖͓͏ ×Ͱด͡Δ
50. खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖ͬͯ
    ֶशϞσϧΛ࡞Ζ͏ʂ ./*45Λ࢖ֶͬͯशϞσϧΛ
    ࡞ΔϓϩάϥϜ
    NOJTUQZΛ
    ࣄલʹೖΕ͓͖ͯ·ͨ͠ͷͰ
    5IPOOZͷ-PBEΛΫϦοΫͯ͠
    ։͖NOJTUQZΛબ୒ͯ͠
    ։͖·͢ɻ
    
51. खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖ͬͯ
    ֶशϞσϧΛ࡞Ζ͏ʂ from keras.datasets import mnist from keras.models import Sequential from keras.layers.core import Dense, Activation from keras.utils import np_utils (X_train, y_train), (X_test, y_test) = mnist.load_data() σʔληοτͷಡΈࠐΈ
    9@USBJO
    
    ֶशͤ͞Δը૾σʔλ
    Z@USBJO
    
    ֶशͤ͞Δϥϕϧσʔλ
    9@UFTU
    
    
    ֶशͤͨ͞ϞσϧΛςετ͢Δςετ༻ը૾σʔλ
    Z@UFTU
    
    
    ֶशͤͨ͞ϞσϧΛςετ͢Δςετ༻ϥϕϧσʔλ NOJTUͷΠϯϙʔτ
52. खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖ͬͯ
    ֶशϞσϧΛ࡞Ζ͏ʂ 9@USBJO
    
    9@USBJOSFTIBQF 
    
    
    
    9@UFTU
    
    9@UFTUSFTIBQF 
    
    
    
    ֶशͤ͞Δը૾σʔλͱςετ༻ը૾σʔλΛ
    ϦγΣΠϓ͠·͢ɻ
    ./*45ͷը૾σʔλ
    ɹɹɹɹɹɹϐΫηϧºϐΫηϧϐΫηϧ
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    ̍ϐΫηϧͷάϨʔεέʔϧ஋Λ̌ʙ̎̑̑ͷ਺ࣈͰදݱ
    
    
    
    
    
    
    
    
    
    
    ɹ
    
    ɹͰׂΔ͜ͱͰ̍ϐΫηϧͷ஋Λ̌ʙ̍ͷؒʹ͠·͢
    ʢਖ਼نԽɿ̌ʙ̍·Ͱ͔͠ѻ͑ͳ͍ͷͰ߹Θͤ·͢ʣ  
53. ./*45ͷը૾ɹɹɹɹը૾Λd·Ͱͷ਺ࣈͰදݱͨ͠ߦྻ
54. खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖ͬͯ
    ֶशϞσϧΛ࡞Ζ͏ʂ Z@USBJO
    
    OQ@VUJMTUP@DBUFHPSJDBM Z@USBJO
    Z@UFTU
    
    OQ@VUJMTUP@DBUFHPSJDBM Z@UFTU
    ,FSBT͸ϥϕϧΛόΠφϦʔͰ͔͠࢖͑ͳ͍ͷͰɺ
    ϥϕϧΛ̌ͱ̍ͷߦྻʹ͢ΔॲཧͰ͢
    
55. खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖ͬͯ
    ֶशϞσϧΛ࡞Ζ͏ʂ NPEFM
    
    4FRVFOUJBM <
    
    
    
    
    
    
    
    
    %FOTF 
    JOQVU@TIBQF 
    
    
    
    
    
    
    
    
    "DUJWBUJPO bTJHNPJE
    
    
    
    
    
    
    
    
    %FOTF 
    
    
    
    
    
    
    
    
    "DUJWBUJPO TPGUNBY
    
    
    
    
    >
    ֶशϞσϧΛ࡞͍ͬͯ·͢ɻ
    
56. खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖ͬͯ
    ֶशϞσϧΛ࡞Ζ͏ʂ NPEFMDPNQJMF MPTTDBUFHPSJDBM@DSPTTFOUSPQZ
    PQUJNJ[FSTHE
    NFUSJDT<BDDVSBDZ>
    NPEFMpU 9@USBJO
    Z@USBJO
    CBUDI@TJ[F
    WFSCPTF
    FQPDIT
    WBMJEBUJPO@TQMJU
    ֶशϞσϧΛ࡞͍ͬͯ·͢ɻ
    
57. खॻ͖਺ࣈͷσʔληοτ./*45Λ࢖ͬͨ
    ֶशϞσϧͷਖ਼౴཰Λग़ྗ TDPSF
    
    NPEFMFWBMVBUF 9@UFTU
    Z@UFTU
    WFSCPTF
    QSJOU UFTU
    BDDVSBDZ
    
    
    TDPSF<>
    
    είΞʢਖ਼౴཰ʣΛग़ྗ͠·͢
    
58. 5IPOOZ 1ZUIPO*%&1ZUIPOͷॳ৺ऀ༻౷߹؀ڥ Λ࢖͏ 3VOϘλϯΛԡ࣮ͯ͠ߦͤ͞Α͏ʂ खॻ͖਺ࣈͷσʔληοτ./*45 &Run mnist2.py Using TensorFlow backend. WARNING: Logging before flag parsing goes to stderr. W0823 09:50:29.745067 1995954896 deprecation_wrapper.py:118] From /home/pi/.local/lib/python3.7/site- packages/tensorflow/__init__.py:98: The name tf.AUTO_REUSE is deprecated. Please use tf.compat.v1.AUTO_REUSE instead. WARNING͕͍͔ͭ͘ग़·͕͢ ໰୊ͳ͍ͷͰ͠͹Β͘଴ͪ·͢
59. 5IPOOZ 1ZUIPO*%&1ZUIPOͷॳ৺ऀ༻౷߹؀ڥ Λ࢖͏ खॻ͖਺ࣈͷσʔληοτ./*45 ਖ਼౴཰͸75.27%
60. վળͯ͠ਖ਼౴཰Λ͋͛Α͏ʂ
61. վળͯ͠ਖ਼౴཰Λ͋͛Α͏ʂ NPEFM
    
    4FRVFOUJBM <
    
    
    
    
    
    
    
    
    %FOTF 
    JOQVU@TIBQF 
    
    
    
    
    
    
    
    
    "DUJWBUJPO bTJHNPJE
    
    
    
    
    
    
    
    
    %FOTF 
    
    
    
    
    
    
    
    
    "DUJWBUJPO TPGUNBY
    
    
    
    
    >
    TJHNPJEΛ
    SFMVʹมߋ͢Δ
    
62. 5IPOOZ 1ZUIPO*%&1ZUIPOͷॳ৺ऀ༻౷߹؀ڥ Λ࢖͏ 3VOϘλϯΛԡ࣮ͯ͠ߦͤ͞Α͏ʂ खॻ͖਺ࣈͷσʔληοτ./*45 &Run mnist2.py Using TensorFlow backend. WARNING: Logging before flag parsing goes to stderr. W0823 09:50:29.745067 1995954896 deprecation_wrapper.py:118] From /home/pi/.local/lib/python3.7/site- packages/tensorflow/__init__.py:98: The name tf.AUTO_REUSE is deprecated. Please use tf.compat.v1.AUTO_REUSE instead. WARNING͕͍͔ͭ͘ग़·͕͢ ໰୊ͳ͍ͷͰ͠͹Β͘଴ͪ·͢
63. NPEFMDPNQJMF MPTTDBUFHPSJDBM@DSPTTFOUSPQZ
    PQUJNJ[FSTHE
    NFUSJDT<BDDVSBDZ>
    NPEFMpU 9@USBJO
    Z@USBJO
    CBUDI@TJ[F
    WFSCPTF
    FQPDIT
    WBMJEBUJPO@TQMJU
    FQPDIT
    ͔ΒFQPDIT
    ʹมߋ͠Α͏
    վળͯ͠ਖ਼౴཰Λ͋͛Α͏ʂ
64. ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ
65. JNQPSU
    OVNQZ
    BT
    OQ
    JNQPSU
    DW
    JNQPSU
    NBUQMPUMJCQZQMPU
    BT
    QMU
    GSPN
    LFSBTEBUBTFUT
    JNQPSU
    NOJTU
    GSPN
    LFSBTNPEFMT
    JNQPSU
    4FRVFOUJBM
    GSPN
    LFSBTMBZFSTDPSF
    
    JNQPSU
    %FOTF
    "DUJWBUJPO
    GSPN
    LFSBTVUJMT
    JNQPSU
    OQ@VUJMT
    GSPN
    LFSBTNPEFMT
    JNQPSU
    MPBE@NPEFM
    
    ൑ผʹඞཁͳ΋ͷΛΠϯϙʔτ͠·͢
    ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ
66. 9@USBJO
    Z@USBJO
    9@UFTU
    Z@UFTU
    
    NOJTUMPBE@EBUB
    9@USBJO
    
    9@USBJOSFTIBQF 
    
    
    
    9@UFTU
    
    9@UFTUSFTIBQF 
    
    
    
    Z@USBJO
    
    OQ@VUJMTUP@DBUFHPSJDBM Z@USBJO
    Z@UFTU
    
    OQ@VUJMTUP@DBUFHPSJDBM Z@UFTU
    NPEFM
    
    4FRVFOUJBM <
    
    
    
    
    
    
    
    
    %FOTF 
    JOQVU@TIBQF 
    
    
    
    
    
    
    
    
    "DUJWBUJPO SFMV
    
    
    
    
    
    
    
    
    %FOTF 
    
    
    
    
    
    
    
    
    "DUJWBUJPO TPGUNBY
    
    
    
    
    >
    NPEFMDPNQJMF MPTTDBUFHPSJDBM@DSPTTFOUSPQZ
    PQUJNJ[FSTHE
    NFUSJDT<BDDVSBDZ>
    NPEFMpU 9@USBJO
    Z@USBJO
    CBUDI@TJ[F
    WFSCPTF
    FQPDIT
    WBMJEBUJPO@TQMJU
    TDPSF
    
    NPEFMFWBMVBUF 9@UFTU
    Z@UFTU
    WFSCPTF
    QSJOU UFTU
    BDDVSBDZ
    
    
    TDPSF<>
    
    ઌ΄ͲͷNOJTUQZͱಉ͡ॲཧͰ
    ਖ਼౴཰Λग़͠·͢ɻ
    ʢ࣌ؒͷؔ܎ͰFQPDITʹͱ͠·͢ʣ
    ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ
67. NPEFMTBWF ./*45I
    JNH
    
    DWJNSFBE KQH
    
    
    ઌ΄ͲͷNOJTUQZͱಉ͡ॲཧͰ
    ࡞ֶͬͨशϞσϧΛอଘ͠·͢ɻ
    ࠓճ͸ࢲ͕ॻ͍ͨ਺ࣈͷ̓ͷը૾Λ
    ೖΕ͓͍ͯͨͷͰɺ
    ͦΕΛ൑ผͯ͠Έ·͠ΐ͏ʂ
    ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ KQH ʻʻ͜ͷը૾͸
    ɹɹJ1IPOFͷϝϞͰॻ͍ͯɺΩϟϓνϟʔͨ͠
    ɹɹࣸਅͷαΠζΛฤूͰτϦϛϯάͨ͠΋ͷͰ͢ɻ
68. HSBZ
    
    DWDWU$PMPS JNH
    DW$0-03@#(3(3":
    DWJNXSJUF HSBZQOH
    HSBZ
    
    PQFODWΛ࢖ͬͯը૾ΛάϨʔεέʔϧʹ͠·͢
    ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ KQH HSBZQOH
69. @
    CJOBSZ
    
    DWUISFTIPME HSBZ
    
    
    DW5)3&4)@#*/"3:
    DWJNXSJUF CJOBSZQOH
    CJOBSZ
    
    PQFODWΛ࢖ͬͯը૾Λ̎஋Խ͠·͢
    ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ CJOBSZQOH HSBZQOH
70. OFHBQPTJ
    
    DWCJUXJTF@OPU CJOBSZ
    DWJNXSJUF OFHBQPTJQOH
    OFHBQPTJ
    
    PQFODWΛ࢖ͬͯը૾ͷനࠇΛ൓స͠·͢
    ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ CJOBSZQOH OFHBQPTJQOH
71. CMVS
    DW(BVTTJBO#MVS OFHBQPTJ  
    
    
    DWJNXSJUF CMVSQOH
    CMVS
    
    PQFODWΛ࢖ͬͯը૾ʹ΅͔͠Λ͍ΕΔ
    ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ OFHBQPTJQOH CMVSQOH
72. PQFODWΛ࢖ͬͯը૾αΠζΛºʹมߋ͢Δ
    ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ CMVSQOH JNH
    
    DWSFTJ[F CMVS 
    
    DW*/5&3@$6#*$
    
    
73. ը૾αΠζΛºʹมߋͨ͠΋ͷΛલॲཧͯ͠
    ઌ΄Ͳ࡞ֶͬͨशϞσϧΛ࢖ͬͯ൑ผ͠
    Ұ൪Մೳੑ͕ߴ͍΋ͷΛදࣔ͢Δɻ ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ
    
    <>
    9BQQFOE JNH
    9
    
    OQBTBSSBZ 9
    9
    
    9
    
    
    9
    
    9SFTIBQF MFO 9
    
    SFTVMU
    
    NPEFMQSFEJDU 9
    QSJOU SFTVMUBSHNBY
    
    
74. 3VOϘλϯΛԡ࣮ͯ͠ߦͤ͞Α͏ʂ &Run mnist2.py Using TensorFlow backend. WARNING: Logging before flag parsing goes to stderr. W0823 09:50:29.745067 1995954896 deprecation_wrapper.py:118] From /home/pi/.local/lib/python3.7/site- packages/tensorflow/__init__.py:98: The name tf.AUTO_REUSE is deprecated. Please use tf.compat.v1.AUTO_REUSE instead. WARNING͕͍͔ͭ͘ग़·͕͢ ໰୊ͳ͍ͷͰ͠͹Β͘଴ͪ·͢ ࣮ࡍʹखͰॻ͍ͨ਺ࣈΛಡΈࠐΜͰ
    ./*45Λ࢖ͬͯ൑ผ͠Α͏ʂ
75. 3BTQCFSSZ1JͰ64#ΧϝϥΛ࢖͏खॱ ࣗ෼ͷखॻ͖਺ࣈΛ൑ผ͠Α͏ʂ
76. 3BTQCFSSZ
    1J
    ͷઃఆΛมߋ͢Δ ϝχϡʔͷɹɹ
    ΛΫϦοΫͯ͠
    ઃఆʼ3TQCFSSZ
    1Jͷઃఆ
    ΛΫϦοΫ͠·͢ɻ
    
77. 3BTQCFSSZ
    1J
    ͷઃఆΛมߋ͢Δ ֤ࣗͷύεϫʔυ ֤ࣗͷύεϫʔυ ɹ
    ⁞3BTQCFSSZ1Jͷઃఆ
    ɹ্ཱ͕͕ͪͬͨΒ
    ɹʮΠϯλʔϑΣΠεʯΛ
    ɹɹબ୒͍ͯͩ͘͠͞ɻ
78. 3BTQCFSSZ
    1J
    ͷઃఆΛมߋ͢Δ ɹΧϝϥΛ༗ޮʹͯ͠
    ɹͯ͠0,Λԡͨ͠Β
    ɹ࠶ىಈ͠·͢ɻ
    
79. 64#Χϝϥͷಈ࡞ςετʢHVWDWJFXΛ࢖͍·͢ʣ TVEP
    BQUHFU
    JOTUBMM
    HVWDWJFX
    ͱೖྗͯ͠Πϯετʔϧ͢Δ
    ్தͰΠϯετʔϧ͕ࢭ·ͬͯ
    ଓߦ͠·͔͢ʁ<:O>
    ͱฉ͔Ε·͢ͷͰ:Λೖྗޙ&OUFSͯ͠
    ଓ͚͍ͯͩ͘͞ɻ
80. Ұ୴Πϯετʔϧ͢Δͱϝχϡʔ͔Β΋ىಈͰ͖·͢ɻ ϝχϡʔͷɹɹ
    ΛΫϦοΫͯ͠
    α΢ϯυͱϏσΦʼHVWDWJFX
    ΛΫϦοΫ͠·͢ɻ
    
81. 64#Χϝϥͷಈ࡞ςετʢHVWDWJFXΛ࢖͍·͢ʣ ΩʔϘʔυͷ*Λԡ͢ͱ
    Ωϟϓνϟʔ͕ࡱΕ·͢ ϝχϡʔ͔ΒͰͳ͘
    λʔϛφϧ͔Β HVWDWJFX
    ͱೖྗͯ͠΋0,Ͱ ͢ɻ
    
82. 64#Χϝϥͷಈ࡞ςετʢHVWDWJFXΛ࢖͍·͢ʣ ίϯτϥε τΛௐ੔͢ Δɻ
    ΩʔϘʔυ ͷ*Λԡ͢ͱ
    Ωϟϓ νϟʔ͕ࡱ Ε·͢
83. 64#Χϝϥͷಈ࡞ςετʢࡱͬͨࣸਅΛ֬ೝ͠·͠ΐ͏ʣ อଘ͞ΕͨࣸਅΛӈΫϦοΫ͠ ͯ
    ϑΝΠϧ໊ͷมߋΛԡͯ͠
    ੩ࢭըͷ໊લΛίϐʔ͠·͠ΐ ͏
    ֬ೝ͠·͠ΐ͏ɻ
    ʢNZ@QIPUP൪߸KQHͰ
    ɹอଘ͞Ε·͢ɻʣ ϑΥϧμϚʔΫΛΫϦοΫͯ͠
    QJϑΥϧμΛ։͖·͠ΐ͏
84. ֶशࡁΈϞσϧΛ࢖ͬͨը૾ೝࣝ
85. ը૾ೝࣝͤ͞Δҝͷ̎ͭͷϑΣʔζ ࣗಈͰը૾ೝࣝΛͤ͞Δʹ͸େ͖͘ೋͭͷϑΣʔζ͕͋Γ·͢ɻ
    ୈ̍ϑΣʔζɿֶशϞσϧΛͭ͘Δ ୈ̎ϑΣʔζɿֶशࡁΈϞσϧΛ࢖͏ σʔλ
    ॲཧ σʔλ
    ऩू ػցֶश
    ਂ૚ֶश ֶशͤ͞Δը૾ ֶशࡁΈ
    Ϟσϧ ֶश༻
    σʔληοτ
    ೝ͍ࣝͤͨ͞ը૾ ೖྗ ग़ྗ ֶशࡁΈ
    Ϟσϧ ਪ࿦ ը૾ೝࣝΛ
    ࢖ͬͨॲཧ
86. ֶशࡁΈϞσϧΛ࢖ͬͯΈΑ͏ʂ ୈ̍ϑΣʔζɿֶशϞσϧΛͭ͘Δ ୈ̎ϑΣʔζɿֶशࡁΈϞσϧΛ࢖͏ σʔλ
    ॲཧ σʔλ
    ऩू ػցֶश
    ਂ૚ֶश ֶशͤ͞Δը૾ ֶशࡁΈ
    Ϟσϧ ֶश༻
    σʔληοτ
    ೝ͍ࣝͤͨ͞ը૾ ೖྗ ग़ྗ ֶशࡁΈ
    Ϟσϧ ਪ࿦ ը૾ೝࣝΛ
    ࢖ͬͨॲཧ
87. ֶशࡁΈϞσϧ*ODFQUJPO7Λ࢖ͬͨը૾ೝࣝʹ௅ઓ
    *ODFQUJPO7ͱ͸ (PPHMF͕։ൃͨ͠σΟʔϓϥʔ χϯάΛ࢖ֶͬͨशϞσϧ
    σʔληοτ*."(&/&5Ͱֶश ͞Ε͍ͯ·͢ɻ
    5FOTPSqPXͰར༻ग़དྷΔ༷ʹҰ ൠެ։͞Ε͍ͯ·͢ɻ
    
88. ݩʹͳΔը૾ɹɹ໿ສຕ
    ϥϕϧͷΧςΰϦ໿ສઍछྨ
    ͷσʔληοτ ֶशࡁΈϞσϧ*ODFQUJPO7Λ࢖ͬͨը૾ೝࣝʹ௅ઓ
    ֶश༻σʔληοτ*NBHF/FUͱ͸͜ͷ෦෼
89. 5IPOOZͷ-PBEΛΫϦοΫͯ͠
    ։͖NPEFMQZΛબ୒ͯ͠
    ։͖·͢ɻ
    ֶशࡁΈϞσϧΛ࢖ͬͯը૾ೝࣝ͢Δ
90. from keras.preprocessing import image from keras.applications.inception_v3 \ import preprocess_input, decode_predictions, InceptionV3 import numpy as np model = InceptionV3(weights='imagenet') img_path=' banana.jpg' img = image.load_img(img_path, target_size=(299,299)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x) preds = model.predict(x) ֶशࡁΈϞσϧΛ࢖ͬͯը૾ೝࣝ͢Δ޻ఔͱ͸ CBOBOBKQH ɹɹɹˣ
    *ODFQUJPO7ͱ
    ͍͏ֶशࡁΈ
    ϞσϧΛ࢖ͬͯ
    ը૾ೝࣝ ɹɹ
    
    
    ˣ
    είΞͰ݁Ռ͕
    ͰΔ
    
91. όφφը૾Ͱ࣮ߦ ̍ʣϝχϡʔͷ36/ϘλϯΛԡ͢ ̎ʣ4IFMMͷʮ3VO
    NPEFMQZʯͱදࣔ͞ΕͯϓϩάϥϜ͕࣮ߦ͞Ε·͢ɻ 8BSOJOH͕ग़·͕͢໰୊ͳ͍Ͱ͢ɻ෼ඵఔ౓Ͱ݁Ռ͕ग़·͢ɻ ̍౓͚ͩΫϦοΫͯ͠ ͍ͩ͘͞ɻԿ౓΋ԡ͞ ͳ͍Α͏ʹɻʢॲཧ͕ ஗͘ͳΓ·͢ɻ
92. from keras.preprocessing import image from keras.applications.inception_v3 \ import preprocess_input, decode_predictions, InceptionV3 import numpy as np model = InceptionV3(weights='imagenet') img_path=' banana.jpg' img = image.load_img(img_path, target_size=(299,299)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x) preds = model.predict(x) print('Predicted:') for p in decode_predictions(preds, top=5)[0]: print("Score {}, Label {}".format(p[2],p[1])) ֶशࡁΈϞσϧΛ࢖ͬͨॲཧͷઆ໌ɿͦͷ̍ ը૾ͷಡΈࠐΈ΍
    ֶशࡁΈϞσϧͷ
    *ODFQUJPO7ͷಡΈ ࠐΈɺ1ZUIPOֶज़ ܭࢉϥΠϒϥϦʔ ͷOVNQZΛΠϯ ϙʔτ͢Δ
93. from keras.preprocessing import image from keras.applications.inception_v3 \ import preprocess_input, decode_predictions, InceptionV3 import numpy as np model = InceptionV3(weights='imagenet') img_path=' banana.jpg' img = image.load_img(img_path, target_size=(299,299)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x) preds = model.predict(x) print('Predicted:') for p in decode_predictions(preds, top=5)[0]: print("Score {}, Label {}".format(p[2],p[1])) ֶशࡁΈϞσϧΛ࢖ͬͨॲཧͷઆ໌ɿͦͷ *NBHF/FUͰֶश͠ ͨॏΈͱ *ODFQUJPO7
    ͷجຊϞσϧ͕ಡ Έࠐ·ΕΔ
94. from keras.preprocessing import image from keras.applications.inception_v3 \ import preprocess_input, decode_predictions, InceptionV3 import numpy as np model = InceptionV3(weights='imagenet') img_path=' banana.jpg' img = image.load_img(img_path, target_size=(299,299)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x) preds = model.predict(x) print('Predicted:') for p in decode_predictions(preds, top=5)[0]: print("Score {}, Label {}".format(p[2],p[1])) ֶशࡁΈϞσϧΛ࢖ͬͨॲཧͷઆ໌ɿͦͷ ೝ͍ࣝͤͨ͞ը૾ ͷࢦఆͱը૾ͷα Πζͷॲཧ
95. from keras.preprocessing import image from keras.applications.inception_v3 \ import preprocess_input, decode_predictions, InceptionV3 import numpy as np model = InceptionV3(weights='imagenet') img_path=' banana.jpg' img = image.load_img(img_path, target_size=(299,299)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x) preds = model.predict(x) print('Predicted:') for p in decode_predictions(preds, top=5)[0]: print("Score {}, Label {}".format(p[2],p[1])) ֶशࡁΈϞσϧΛ࢖ͬͨॲཧͷઆ໌ɿͦͷ ը૾ͷલॲཧ
    
96. from keras.preprocessing import image from keras.applications.inception_v3 \ import preprocess_input, decode_predictions, InceptionV3 import numpy as np model = InceptionV3(weights='imagenet') img_path=' banana.jpg' img = image.load_img(img_path, target_size=(299,299)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x) preds = model.predict(x) print('Predicted:') for p in decode_predictions(preds, top=5)[0]: print("Score {}, Label {}".format(p[2],p[1])) ֶशࡁΈϞσϧΛ࢖ͬͨॲཧͷઆ໌ɿͦͷ ਪ࿦
97. from keras.preprocessing import image from keras.applications.inception_v3 \ import preprocess_input, decode_predictions, InceptionV3 import numpy as np model = InceptionV3(weights='imagenet') img_path=' banana.jpg' img = image.load_img(img_path, target_size=(299,299)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x) preds = model.predict(x) print('Predicted:') for p in decode_predictions(preds, top=5)[0]: print("Score {}, Label {}".format(p[2],p[1])) ֶशࡁΈϞσϧΛ࢖ͬͨॲཧͷઆ໌ɿͦͷ ਪ࿦݁Ռͷग़ྗ
98. όφφͷ݁Ռ UIPOOZͱ͍͏σϑΥϧτͰ
    ೖ͍ͬͯΔQZUIPOͷ*%&Λ
    ࠓճ࢖͍·͢ɻ
    LFSBTΛ࢖ͬͨը૾ೝࣝΛ
    ࢼ͢ͷʹTBNQMFQZΛ
    ։͖·͢ɻ
    
    DE
    d
    
    UIPOOZ
    TBNQMFQZ
    ͱλʔϛφϧͰଧ͍ͬͯͩ͘͞ɻ
    είΞ
    όφφͱೝࣝ
99. from keras.preprocessing import image from keras.applications.inception_v3 \ import preprocess_input, decode_predictions, InceptionV3 import numpy as np model = InceptionV3(weights='imagenet') img_path=' apple.jpg' img = image.load_img(img_path, target_size=(299,299)) x = image.img_to_array(img) x = np.expand_dims(x, axis=0) x = preprocess_input(x) preds = model.predict(x) ֶशࡁΈϞσϧΛ࢖ͬͯը૾ೝࣝ͢Δ޻ఔͱ͸ BQQMFKQH ɹˣ
    ɹߦ໨ͷ
    CBOBOBKQHΛ
    ɹBQQMFKQHʹ
    มߋ͢Δ ɹɹ
    
    
    ˣ
    ಉ͡Α͏ʹ࣮ߦ
    ͯ݁͠ՌΛΈΔ
    
100. ΓΜ͝ը૾Ͱ࣮ߦ ̍ʣϝχϡʔͷ36/ϘλϯΛԡ͢ ̎ʣ4IFMMͷʮ3VO
     NPEFMQZʯͱදࣔ͞ΕͯϓϩάϥϜ͕࣮ߦ͞Ε·͢ɻ ಉ͡Α͏ʹ̎෼̏̌ඵఔͰ݁Ռ͕ग़·͢ɻͲͷΑ͏ͳ݁ՌʹͳΔͰ͠ΐ͏͔ʁʁ
101. ݁Ռ͕ग़Δ·Ͱ̑෼΄Ͳٳܜ
102. ΓΜ͝ͷ݁Ռ είΞ
     (SBOOZ@4NJUIͱೝࣝ ͋Εʁ
     BQQMF͡Όͳ͍ͷʁʁ (SBOOZ@4NJUIͱ͸ΓΜ͝ͷ඼छͰ͢ɻֶशͤ͞Δ࣌ͷϥϕϧ͕݁ՌʹͰ·͢ɻ
103. σʔληοτͷ඼࣭ͱ༧ଌ݁Ռ (SBOOZ
     4NJUI ͜Ε͸౰વ
     BQQMFͱ
     ༧ଌ͞ΕΔ
     ΑͶ ɹσʔληοτͰॻ͔Ε͍ͯΔϥϕϧ ˣ
     ಛ௃͕Ұ൪ࣅ͍ͯΔͱ൑அ͞ΕΔ
     ˣ
     ༧ଌ݁Ռʮ(SBOOZ
     4NJUIʯ σʔληοτͷ
     ඼࣭͸
     ͱͯ΋ॏཁ
104. ࣮ࡍʹΧϝϥͰࡱͬͨ΋ͷΛೝࣝͯ͠Έ·͠ΐ͏
     Ұ୴Πϯετʔϧ͢Δͱϝχϡʔ͔Β΋ىಈͰ͖·͢ɻ ϝχϡʔͷɹɹ
     ΛΫϦοΫͯ͠
     α΢ϯυͱϏσΦʼHVWDWJFX
     ΛΫϦοΫ͠·͢ɻ
     
105. 64#Χϝϥͷಈ࡞ςετʢࡱͬͨࣸਅΛ֬ೝ͠·͠ΐ͏ʣ อଘ͞ΕͨࣸਅΛΫϦοΫͯ͠
     ੩ࢭըΛ֬ೝ͠·͠ΐ͏ɻ
     ʢNZ@QIPUP൪߸KQHͰ
     ɹอଘ͞Ε·͢ɻʣ ϑΥϧμϚʔΫΛΫϦοΫͯ͠
     QJϑΥϧμΛ։͖·͠ΐ͏ ΩʔϘʔυͷ*Λԡ͢ͱ
     Ωϟϓνϟʔ͕ࡱΕ·͢
106. ࠓΧϝϥͰࡱͬͨը૾Ͱࢼͯ͠ΈΑ͏ɿ ࡱͬͨը૾ ɹˣ
     ɹߦ໨ͷ
     BQQMFKQHΛ
     ɹը૾໊ʹ
     มߋ͢Δ ɹɹ
     
     
     ˣ
     ಉ͡Α͏ʹ࣮ߦ
     ͯ݁͠ՌΛΈΔ
     ‘ࡱͬͨը૾໊’
107. ΧϝϥͰࡱͬͨը૾Ͱ࣮ߦ ̍ʣϝχϡʔͷ36/ϘλϯΛԡ͢ ̎ʣ4IFMMͷʮ3VO
     NPEFMQZʯͱදࣔ͞ΕͯϓϩάϥϜ͕࣮ߦ͞Ε·͢ɻ ಉ͡Α͏ʹ̑෼ඵఔ౓Ͱ݁Ռ͕ग़·͢ɻͲͷΑ͏ͳ݁ՌʹͳΔͰ͠ΐ͏͔ʁʁ
108. 3BTQCFSSZ1JͳΒ͜Μͳ͜ͱग़དྷ·͢ʂ
     ΈΜͳͷϥζύΠίϯςετͰ༏ྑ৆ɹ࡞඼ྫʣ
109. Microsoft Azure Λ࢖༻ͯ͠ সإ൑ఆ Emotion API ௒খܕΧϝϥ RaspberryPi ΧϝϥͰ ࡱӨͨ͠ը૾ 3BTQCFSSZ1JͱػցֶशΛ࢖ͬͨྫɹɹɹɹ
     3BTQCFSSZ1JΛ࢖ͬͨ࡞඼ΛԠืͯ͠ΈΑ͏ʂɹɹɹ
110. ΧϝϥͰࡱͬͨը૾ͷ݁Ռ Έͳ͞Μ͸ԿΛ
     ࡱӨ͠·͔ͨ͠ʁ
     ݁ՌΛͥͻڭ͑ͯ
     ͍ͩ͘͞
111. ωοτͰݕࡧͨ͠ը૾Ͱࢼͯ͠ΈΑ͏
     Έͳ͞Μ͸ԿΛ
     ࢼ͠·͔ͨ͠ʁ
     ݁ՌΛͥͻڭ͑ͯ
     ͍ͩ͘͞ $ISPNJVNͱ͍͏ϒϥ΢β্ཱ͕͕ͪΔͷͰωοτݕࡧͯ͠Έ͍ͯͩ͘͞ɻ
112. ࠓ೔Έͳ͞Μ͸ֶशϞσϧͷ࡞੒ͱ
     ࡞੒͞ΕֶͨशϞσϧͰࣗ෼͕ࡱӨͨ͠
     ը૾Λ࢖͍ը૾ೝࣝͤ͞Δ͜ͱ͕Ͱ͖·ͨ͠ɻ
     ͜ΕͰແࣄ3BTQCFSSZ1JͰσΟʔϓϥʔχϯάΛ
     ͸͡ΊΔ͜ͱ͕Ͱ͖·ͨ͠ʂ ͓ΊͰͱ͏͍͟͝·͢
113. PQFODWΛ࢖ͬͨإೝࣝ
114. 5IPOOZͷ-PBEΛΫϦοΫͯ͠
     ։͖GBDFQZΛબ୒ͯ͠
     ։͖·͢ɻ
     PQFODWΛ࢖ͬͯإೝࣝΛ͢Δલʹ
115. ࣮ߦ ̍ʣϝχϡʔͷ36/ϘλϯΛԡ͢ ̎ʣΧϝϥը૾ͱάϨΠεέʔϧʹมߋͤͨ͞ը૾্ཱ͕͕ͪΓ·͢ ̏ʣࣗ෼ͷإ͕ݟ͑ΔҐஔʹΧϝϥΛஔ͍ͯ֬ೝͰ͖ͨΒ4501Λԡ͢
116. إೝࣝ͢Δલʹը૾ͷ൓సΛ௚ͦ͏ SFU
     GSBNF
     
     DBQSFBE
     
     GSBNF
     
     DWqJQ GSBNF
     
     
     HSBZ
     
     DWDWU$PMPS GSBNF
     DW$0-03@#(3(3": Λ̍ʹ͠·͢ɻ
     
117. ΋͏Ұ౓࣮ߦ ̍ʣϝχϡʔͷ36/ϘλϯΛԡ͢ ̎ʣ൓స͕௚ͬͨը૾͕ݟ͑Δ͸ͣɻ ̏ʣ֬ೝͰ͖ͨΒ4501Λԡ͢
118. 5IPOOZͷ-PBEΛΫϦοΫͯ͠
     ։͖GBDF̎QZΛબ୒ͯ͠
     ։͖·͢ɻ
     PQFODWΛ࢖ͬͯإೝࣝΛ͢Δલʹ
119. ࣮ߦ ̍ʣϝχϡʔͷ36/ϘλϯΛԡ͢ ̎ʣإͷྠֲͱ໨Λೝࣝ͠·͢ɻ
120. Χεέʔυ෼ྨث ࣗ෼ͷإ͕ແࣄೝࣝͰ͖ͨΒFTDΩʔΛԡ͢ɻ΋͘͠͸4501
     ͳʹ͔Λೝࣝ͢Δʹ͸ɺͦͷಛ௃Λநग़ͨ͠σʔλ͕ඞཁͰ
     ࠓճ࢖͍ͬͯΔͷ͕ʮΧεέʔυ෼ྨثʯͰ͢ɻ
     
121. ⭐ਖ਼౴཰͸·ͩ·ׂ͙ͩ̕Β͍ͳͷͰɺਫ਼౓Λ͋͛Δ ࠓޙ͸͜ͷߨ࠲Λडߨͨ͠ܦݧΛੜ͔ͯ͠
     ⭐ϦΞϧλΠϜͰ਺ࣈΛೝࣝͰ͖ΔΑ͏ʹ͢Δ ⭐ࣗ෼ͰσʔληοτΛ࡞ͬͯΈΔɹɹɹͳͲͳͲ
122. ߨٛ͸͜ΕͰऴྃͰ͢ɻ
     ͝੩ௌ͋Γ͕ͱ͏͍͟͝·ͨ͠ɻ