infoShare AI Roadshow 2018 - Krzysztof Kudryński & Błażej Kubiak (TomTom) - Deep learning in computer vision
Download as PPSX, PDF•1 like•118 views
The document outlines a plan for a presentation on deep learning in computer vision by Krzysztof Kudryński and Błażej Kubiak, covering fundamental concepts such as convolutional neural networks (CNNs) and their applications through a project walkthrough. It highlights various challenges in training deep networks, including handling issues like the vanishing gradient and the need for smart initialization and regularization techniques. The document includes references to multiple video lectures and training methodologies related to the topic.
![[n, 200, 180, 4][5, 5, 3, 4]
n
[n, 200, 180, 3]
200
180
n
3
4
3
5
5
4
200
180
DeepLearninginComputerVision
KrzysztofKudryński](https://image.slidesharecdn.com/poznan-wroclaw-warszawa-deeplearningincomputervisioninfoshare-181206144228/85/infoShare-AI-Roadshow-2018-Krzysztof-Kudrynski-Blazej-Kubiak-TomTom-Deep-learning-in-computer-vision-20-320.jpg)
![vertical
edges
horizon
edges
frame
glasses
23
345
3%
97%
[n, 200, 180, 1] [3, 3, 1, 4] [n, 200, 180, 4]
[n, 20, 18, 4] [5, 5, 4, 1] [n, 20, 18, 1]
[n, 10, 9, 1] [6, 6, 1, 1] [n, 10, 9, 1]
[n, 5, 5, 1]
[25, 2] [n, 2] [n, 2]
placeholder
session
Parse
images with
labels
[0 1]
[0 1]
[1 0]
[1 0]
Deep Learning in Computer Vision
Krzysztof Kudryński](https://image.slidesharecdn.com/poznan-wroclaw-warszawa-deeplearningincomputervisioninfoshare-181206144228/85/infoShare-AI-Roadshow-2018-Krzysztof-Kudrynski-Blazej-Kubiak-TomTom-Deep-learning-in-computer-vision-21-320.jpg)
![model = Sequential()
model.add(Dense(rowSize, input_dim=rowSize, activation='relu'))
model.add(Dense(rowSize, activation='relu'))
model.add(Dense(1, activation='sigmoid'))
adam = optimizers.adam()
model.compile(loss='binary_crossentropy',
optimizer=adam,
metrics=['accuracy'])
model.fit_generator(generate_arrays_from_file(trainArr, tranLabelMap, 30),
steps_per_epoch=2000, epochs=100,
validation_data=(valData, valDataLabels))
Training Network in Keras
Deep Learning in Computer Vision
Krzysztof Kudryński](https://image.slidesharecdn.com/poznan-wroclaw-warszawa-deeplearningincomputervisioninfoshare-181206144228/85/infoShare-AI-Roadshow-2018-Krzysztof-Kudrynski-Blazej-Kubiak-TomTom-Deep-learning-in-computer-vision-62-320.jpg)
infoShare AI Roadshow 2018 - Krzysztof Kudryński & Błażej Kubiak (TomTom) - Deep learning in computer vision
- 1. KRZYSZTOF KUDRYŃSKI BŁAŻEJ KUBIAK DEEP LEARNING IN COMPUTER VISION
- 2. Plan • Get excited (7 min) about the world we live in • Understand CNN (18 min) from fundamentals to state-of-the art approach • Apply CNN (15 min) by walk-through a challenging computer-vision project. • Get scared (1 min) about the world we’ll live in Deep Learning in Computer Vision Krzysztof Kudryński
- 3. (1982) www.youtube.com/watch?v=qHepKd38pr0Deep Learning in Computer Vision Krzysztof Kudryński
- 4. Deep Learning in Computer Vision Krzysztof Kudryński
- 5. (2017) Deep Learning in Computer Vision Krzysztof Kudryński
- 6. (2017) www.youtube.com/watch?v=W6wzUMk9w-cDeep Learning in Computer Vision Krzysztof Kudryński
- 7. (2008) Deep Learning in Computer Vision Krzysztof Kudryński
- 8. (2016) www.youtube.com/watch?v=FKXOucXB4a8Deep Learning in Computer Vision ⓒKrzysztof Kudryński
- 9. www.youtube.com (1982) Deep Learning in Computer Vision Krzysztof Kudryński
- 10. (2015) Deep Learning in Computer Vision Krzysztof Kudryński
- 11. (2015) Deep Learning in Computer Vision Krzysztof Kudryński
- 12. (2015) Deep Learning in Computer Vision Krzysztof Kudryński
- 13. A small bird with yellow breast, brown crown and black superciliary (2015) Deep Learning in Computer Vision Krzysztof Kudryński
- 14. Deep Learning in Computer Vision Krzysztof Kudryński
- 15. (2015-2018) Deep Learning in Computer Vision Krzysztof Kudryński
- 16. Deep Learning in Computer Vision Krzysztof Kudryński
- 17. Deep Learning in Computer Vision Krzysztof Kudryński
- 18. 100% 0% 0% Deep Learning in Computer Vision Krzysztof Kudryński
- 19. vertical edges horizon edges frame glasses 23 345 3% 97% Convolution Max-pooling Deep Learning in Computer Vision Krzysztof Kudryński
- 20. [n, 200, 180, 4][5, 5, 3, 4] n [n, 200, 180, 3] 200 180 n 3 4 3 5 5 4 200 180 DeepLearninginComputerVision KrzysztofKudryński
- 21. vertical edges horizon edges frame glasses 23 345 3% 97% [n, 200, 180, 1] [3, 3, 1, 4] [n, 200, 180, 4] [n, 20, 18, 4] [5, 5, 4, 1] [n, 20, 18, 1] [n, 10, 9, 1] [6, 6, 1, 1] [n, 10, 9, 1] [n, 5, 5, 1] [25, 2] [n, 2] [n, 2] placeholder session Parse images with labels [0 1] [0 1] [1 0] [1 0] Deep Learning in Computer Vision Krzysztof Kudryński
- 22. 0 Deep Learning in Computer Vision Krzysztof Kudryński
- 23. 97% glasses99% glasses 51% glasses DeepLearninginComputerVision KrzysztofKudryński
- 24. 0a Deep Learning in Computer Vision Krzysztof Kudryński
- 25. 0b Deep Learning in Computer Vision Krzysztof Kudryński
- 26. 1 Deep Learning in Computer Vision Krzysztof Kudryński
- 27. 1 Deep Learning in Computer Vision Krzysztof Kudryński
- 28. Citations: 3000 ALWAYS POSITIVE Deep Learning in Computer Vision Krzysztof Kudryński
- 29. 1b Deep Learning in Computer Vision Krzysztof Kudryński
- 30. 1c Deep Learning in Computer Vision Krzysztof Kudryński
- 31. 2 Deep Learning in Computer Vision Krzysztof Kudryński
- 32. 2 Deep Learning in Computer Vision Krzysztof Kudryński
- 33. 3 Deep Learning in Computer Vision Krzysztof Kudryński
- 34. 4 LADY #0 LADY #1 NO GLASSES HAS GLASSES Deep Learning in Computer Vision Krzysztof Kudryński
- 35. 6a (5) correct correct correct correct correct wrong wrong Deep Learning in Computer Vision Krzysztof Kudryński
- 36. 6b (6) Deep Learning in Computer Vision Krzysztof Kudryński
- 37. 7a Deep Learning in Computer Vision Krzysztof Kudryński
- 38. Deep Learning in Computer Vision Krzysztof Kudryński
- 39. ) Deep Learning in Computer Vision Krzysztof Kudryński
- 40. 7b Deep Learning in Computer Vision Krzysztof Kudryński
- 41. We need to cut the costs! Optimizer(). minimize(cost) Deep Learning in Computer Vision Krzysztof Kudryński
- 42. We need to cut the costs! Optimizer(). minimize(cost) Deep Learning in Computer Vision Krzysztof Kudryński
- 43. ) We need to cut the costs! Optimizer(). minimize(cost) Deep Learning in Computer Vision Krzysztof Kudryński
- 44. Deep Learning in Computer Vision Krzysztof Kudryński
- 45. 7e Deep Learning in Computer Vision Krzysztof Kudryński
- 46. 8b Deep Learning in Computer Vision Krzysztof Kudryński
- 47. • Smart initialization • Regularization • Gradient clipping • New activation functions • Huge training sets • Deep networks Sophisticated problems Clever solutions • Hard to train • Vanishing gradient • Exploding gradient Deep Learning in Computer Vision Krzysztof Kudryński
- 48. • Smart initialization • Regularization • Gradient clipping • New activation functions • Intermediate training paths • Huge training sets • Deep networks • Hard to train • Vanishing gradient • Exploding gradient Sophisticated problems Clever solutions Deep Learning in Computer Vision Krzysztof Kudryński
- 49. • Smart initialization • Regularization • Gradient clipping • New activation functions • Intermediate training paths • Residual connections • Huge training sets • Deep networks • Hard to train • Vanishing gradient • Exploding gradient Sophisticated problems Clever solutions Deep Learning in Computer Vision Krzysztof Kudryński
- 50. Deep Learning in Computer Vision Krzysztof Kudryński
- 51. POLAND /112018 Deep Learning in Computer Vision Krzysztof Kudryński
- 52. Deep Learning in Computer Vision Krzysztof Kudryński
- 53. Deep Learning in Computer Vision Krzysztof Kudryński
- 54. Deep Learning in Computer Vision Krzysztof Kudryński
- 55. RGB Image Convolution Max pooling X X Convolution Max pooling A B C ||A – B|| ||B – C|| L = ||A – Ex||2 L = ||B – C||2 – ||A – B||2 Deep Learning in Computer Vision Krzysztof Kudryński
- 56. Deep Learning in Computer Vision Krzysztof Kudryński
- 57. Convolution Max pooling X X Convolution Max pooling A B |A – B| Face recognition network Deep Learning in Computer Vision Krzysztof Kudryński
- 58. Deep Learning in Computer Vision Krzysztof Kudryński
- 59. Kate Winslet Arnold Schwarzenegger Alec Baldwin Abdullah Deep Learning in Computer Vision Krzysztof Kudryński
- 60. Convolution Max pooling X X Convolution Max pooling A B |A – B| Face recognition network 1 Deep Learning in Computer Vision Krzysztof Kudryński
- 61. Convolution Max pooling X X Convolution Max pooling A B |A – B| Face recognition network 0 Deep Learning in Computer Vision Krzysztof Kudryński
- 62. model = Sequential() model.add(Dense(rowSize, input_dim=rowSize, activation='relu')) model.add(Dense(rowSize, activation='relu')) model.add(Dense(1, activation='sigmoid')) adam = optimizers.adam() model.compile(loss='binary_crossentropy', optimizer=adam, metrics=['accuracy']) model.fit_generator(generate_arrays_from_file(trainArr, tranLabelMap, 30), steps_per_epoch=2000, epochs=100, validation_data=(valData, valDataLabels)) Training Network in Keras Deep Learning in Computer Vision Krzysztof Kudryński
- 63. (2018) source:http://dx.doi.org/10.1101/272518 Deep Learning in Computer Vision Krzysztof Kudryński
- 64. (2018) source:http://dx.doi.org/10.1101/272518 Deep Learning in Computer Vision Krzysztof Kudryński
- 65. Krzysztof Kudrynski [email protected] @kriskudrynski Blazej Kubiak [email protected] @blazkubiak Deep Learning in Computer Vision Krzysztof Kudryński