deep learning for image recognition in python
DESCRIPTION
"Deep Learning for Image Recognition in Python" at PyCon JP 2014 https://pycon.jp/2014/schedule/presentation/20/ Youtube http://youtu.be/JWGXQhVHTTA Keywords Machine Learning, Object Recognition, Face Recognition, Artificial Intelligence (AI) ディープラーニング, 深層学習, 機械学習, 画像認識, 物体認識, 顔認識, 人工知能TRANSCRIPT
- 1. Deep Learning for Image Recognition in Python x Hideki Tanaka PyCon JP 2014
- 2. PyCon?
- 3. PyCon #pyconjp
- 4. @atelierhide
- 5. @atelierhide = ?
- 6. @atelierhide = Lens Designer
- 7. @atelierhide = Photographer
- 8. @atelierhide = Pythonista
- 9. @atelierhide = Kaggler
- 10.
- 11. Image Recognition x Deep Learning
- 12. x
- 13. (AI)
- 14. (AI)
- 15. Pepper
- 16. 1
- 17. Pepper?
- 18. Agenda 1. Image Recognition? 2. Deep Learning? 3. Pepper
- 19. Image Recognition?
- 20. Dog or Cat?
- 21. ?
- 22. (scikit-learn)
- 23. SURF Haar-like features Python OpenCV Mahotas
- 24. Training set Test set : 25,000 images : 12,500 images
- 25. Accuracy 60% with Haar-like features
- 26.
- 27. Accuracy >95% with Deep Learning
- 28. 98.5%!?
- 29. Deep Learning
- 30. Deep Learning?
- 31. input output hidden n
- 32. Core Language Binding Theano/Pylearn2 cuda-convnet OverFeat Caffe DeCAF Python C++ Lua C++ Python - Python Python Python - Pre-trained Networks Framework
- 33. Pre-trained Networks ImageNet Large Scale Visual Recognition Challenge Deep Learning Deep Convolutional Neural Networks
- 34. Pre-trained Networks (DeCAF) (scikit-learn)
- 35. Accuracy 96% with Deep Learning
- 36.
- 37. Deep Learning Pepper
- 38. Pepper?
- 39. Pepper
- 40. Pepper
- 41. ! !
- 42.
- 43. JKC48
- 44. JKC48 48
- 45. JKC48
- 46. 1. 2. 3. Deep Learning 4. Pepper 5. Pepper
- 47.
- 48. 1. 2. 3. Deep Learning 4. Pepper 5. Pepper
- 49. Demo
- 50. JKC48
- 51. Ubuntu 12.04 pyenv + miniconda3.4.2 (Python 2.7.8) Flask, OpenCV, DeCAF
- 52. #pyconjp
- 53.
- 54. One More Thing
- 55. !
- 56.
- 57.
- 58.
- 59. The esperanto generator torajiro aida 11:30- Media Hall
- 60. !