オフィスの前にある信号が変わる タイミング教えてくれるWebページ 作ろうとしたよ with DeepLearning

2017年末に社内で開かれたハッカソンの発表資料を社外向けに少し修正したものです。

Transcript

1. ΦϑΟεͷલʹ͋Δ৴߸͕มΘΔ
 λΠϛϯάڭ͑ͯ͘ΕΔ Webϖʔδ
 ࡞ͬͨΑ࡞Ζ͏ͱͨ͠Α Hackday2017 Team4 ञҪ ਸࢸ ※Hackday2017ͱ͸ɺ౦ূҰ෦্৔اۀͷגࣜձࣾϑΝϯίϛϡχέʔγϣϯζࣾ಺Ͱ೥຤ͷ Ջͳ༨༟ͷ͋Δ࣌ʹߦΘΕͨνʔϜ੍ϋοΧιϯͷ͜ͱͰ͢
2. ੒Ռ෺ ͜Μͳײ͡
3. ͳͥ࡞͔ͬͨ - ΦϑΟεͷલͷาߦऀ৴߸ͷ଴ͪ࣌ؒ݁ߏ ௕͍ - ੨ʹͳΔ·Ͱͷ͕࣌ؒ෼͔Ε͹ɺ੮Λཱͭ λΠϛϯά΋෼͔Δ͸ͣ
4. ΈΜͳϋοϐʔ
 ؒҧ͍ͳ͍ʂ - ΦϑΟεͷલͷาߦऀ৴߸ͷ଴ͪ࣌ؒ݁ߏ ௕͍ - ੨ʹͳΔ·Ͱͷ͕࣌ؒ෼͔Ε͹ɺ੮Λཱͭ λΠϛϯά΋෼͔Δ͸ͣ
5. ֓ཁ - ৴߸ͷมΘΔपظ͸༧Ίଌ͓ͬͯ͘ - ͨ·ʹը૾ೝࣝͰ੺੨੾ΓସΘΓ
 λΠϛϯάΛิਖ਼͢Δ
6. पظ؅ཧαʔό पظऔಘ ৴߸ͷ৭ ৴߸ͷपظΛ؅ཧ ੺੨൑ఆϓϩάϥϜ શମߏ੒ ৴߸ͷը૾ࡱӨ ৴߸ͷ৭Λ൑ఆ ϒϥ΢β ৴߸ͷλΠϛϯάΛදࣔ
7. ৴߸ͷ੺੨ೝࣝͷ
 ͨΊʹ΍ͬͨ͜ͱ
8. ͜ΜͳΧϝϥͰ
9. ͜Μͳը૾ͷ৴߸ͷ৭Λ
10. ͜͜ʹ͋Δʢ੨ʣ
11. ൑ఆ͍ͨ͠ʂ
12. ͜͏͍͏ը૾ॲཧͱ͍͑͹
13. Deep Learning Ͱ͢ΑͶ…
14. ཁ݅ - WebΧϝϥͰࡱͬͨը૾Λ࢖͏ - ҎԼ͸ېࢭ - खಈͰ৴߸ʹζʔϜ - खಈͰը૾Ճ޻ - ΧϝϥΛ׬શʹݻఆ͢Δ
15. ·ͣ͸ֶशσʔλ࡞Γ
 ʢ৭Μͳ֯౓͔ΒࡱΔ,໿5000ຕʣ ੺ ੨ ੺ ੨
16. PythonͷίʔυΛΨʔοͱॻ͍ͯ ʢ200ߦ͘Β͍ʣ … def vgg_std16_model(img_rows, img_cols): model = Sequential() model.add(ZeroPadding2D((1, 1), input_shape=(3, img_rows, img_cols))) model.add(Convolution2D(64, 3, 3, activation='relu')) model.add(ZeroPadding2D((1, 1))) model.add(Convolution2D(64, 3, 3, activation='relu')) model.add(MaxPooling2D((2, 2), strides=(2, 2))) model.add(ZeroPadding2D((1, 1))) model.add(Convolution2D(128, 3, 3, activation='relu'))
17. ֶशʂ(ŕŦŖƃ ~/hackday/python$ python3 train_and_evaluate.py 4591 train samples Start training........... Train on 3121 samples, validate on 551 samples 10/3121 [=>..............................] - ETA: 35023s - loss: 0.5735 - acc: 0.6032
18. Μ…ʁ ~/hackday/python$ python3 train_and_evaluate.py 4591 train samples Start training........... Train on 3121 samples, validate on 551 samples 10/3121 [=>..............................] - ETA: 35023s - loss: 0.5735 - acc: 0.6032
19. ࢒Γ࣌ؒ35023s ≒ 10࣌ؒ ~/hackday/python$ python3 train_and_evaluate.py 4591 train samples Start training........... Train on 3121 samples, validate on 551 samples 10/3121 [=>..............................] - ETA: 35023s - loss: 0.5735 - acc: 0.6032
20. ๻ʮऴΘΒͳ͍… Ͳ͏͢Ε͹…ʯ
21. ʁʮCPU͕΍ΒΕͨΑ͏ͩͳ…ʯ
22. ๻ʮ͋ɺ͋ͳͨ͸…ʂʂʯ
23. ๻ʮGPU͞Μʂʯ
24. ͬͯ͜ͱͰGPUͰ࠶ֶशʂ(ŕŦŖƃ ※AWSͷGPUΠϯελϯε࢖͍·ͨ͠ ~/hackday/python$ python3 train_and_evaluate.py Using gpu device 0: Tesla M60 (CNMeM is disabled, cuDNN 4007) 4591 train samples Start training........... Train on 3121 samples, validate on 551 samples 10/3121 [=>..............................] - ETA: 2714s - loss: 0.5735 - acc: 0.6032
25. 1࣌ؒҎ಺ͰऴΘΔʂ ~/hackday/python$ python3 train_and_evaluate.py Using gpu device 0: Tesla M60 (CNMeM is disabled, cuDNN 4007) 4591 train samples Start training........... Train on 3121 samples, validate on 551 samples 10/3121 [=>..............................] - ETA: 2714s - loss: 0.5735 - acc: 0.6032
26. 1࣌ؒޙ…
27. ๻ʮ͓ɺֶशऴΘͬͯΔ…ʯ 3121/3121 [==============================] - 0s - loss: 0.0051 - acc: 1.0000 - val_loss: 0.0085 - val_acc: 1.0000
28. ๻ʮਫ਼౓͸… 100%ʂʁʯ 3121/3121 [==============================] - 0s - loss: 0.0051 - acc: 1.0000 - val_loss: 0.0085 - val_acc: 1.0000
29. ๻ʮਫ਼౓͸… 100%ʂʁʯ
 ๻ʮ͜ͷউෛ΋ΖͨͰʂʯ 3121/3121 [==============================] - 0s - loss: 0.0051 - acc: 1.0000 - val_loss: 0.0085 - val_acc: 1.0000
30. 1೔໨ऴྃ
31. 2೔໨
32. ๻ʮͯ͞ϦΞϧλΠϜʹࡱͬ ͨ৴߸ͷ৭Λ༧ଌ͢Δ͔…ʯ
33. PCʮ੨ʂʯ ๻ʮਖ਼ղʂʯ
34. PCʮ੺ʂʯ ๻ʮ͍͢͝ʂʯ
35. PCʮ੺ʂʯ ๻ʮ͋Ε…ʁʯ
36. PCʮ੨ʂʯ ๻ʮΜΜΜ...ʁʯ
37. ๻ʮ͍ͭ͜΋͠΍…ʯ
38. ๻ʮԣஅาಓͷ্ʹਓ͕͍Δ͔Ͳ ͏͔Ͱ൑அͯ͠΍͕Δʂʂʂʯ
39. Deep Learning͸෺ମೝࣝೳྗ͕
 ߴ͗ͯ͢ɺਓؒͰ͸ࢥ͍͔ͭͳ͍Α͏ ͳϧʔϧΛউखʹ࡞ͬͯ͠·͏ͷͰ͢ɻ
40. ๻ʮͰ΋͜Ε͸๻͕ࡱͬͨσʔλ ʹภΓ͕͚͋ͬͨͩ…ʯ
41. ๻ʮҎԼͷΑ͏ͳը૾Λͨ͘͞Μ ࡱͬͯ࠶ֶश΍ʂʯ - ੺͚ͩͲ౉ͬͯΔਓ͕͍Δࣸਅ - ੨͚ͩͲ୭΋౉ͬͯͳ͍ࣸਅ
42. ࠶ֶशޙ…
43. PCʮ੺ʂʯ ๻ʮΑ͠Α͠ʯ
44. PCʮ੨ʂʯ ๻ʮ͓΍ʁʯ
45. ๻ʮ͍ͭ͜…ʯ
46. ๻ʮࠓ౓͸͜͜Λं͕૸ͬͯΔ͔ Ͳ͏͔Ͱ൑ఆͯ͠΍͕Δʂʯ
47. ҎԼ͍ͨͪͬ͜͝ʢഊ๺ʣ
48. ݁Ռ - ࠷ऴతʹ·͊·͊ͳਫ਼౓ʹ͸ͳͬͨ
 ʢϦΞϧλΠϜը૾Ͱ90%͘Β͍ʁʣ - Ͱ΋ɺࠓճͷ໨తͷͨΊʹ͸ਫ਼౓ෆ଍ - ภΓͷͳֶ͍शσʔλΛ΋ͬͱͨ͘͞Μ
 ࡱΕΕ͹ղܾ͢Δ͸ͣ
49. ݸਓతײ૝ - Deep Learning͍͢͝
 - GPU͍͢͝
 - ྑֶ͍शσʔλΛ࡞ΔͷΉ͍ͣ