![Basic MATLAB commands
• net = Alexnet % network ojects
• Lyers = net.Layers
• Input layer = Lyers(1) % first layer input layer
• ClassLayer = ly(end) % is the class name
• CategoryNames = ClassLayer.ClassNames
• [pred, percent] = classify(net, ImagePointer)
31](https://image.slidesharecdn.com/introductiontodnn-180307173404/85/Brief-Introduction-to-Deep-Learning-for-Object-Recognition-Using-MATLAB-31-320.jpg)
Brief Introduction to Deep Learning for Object Recognition Using MATLAB
- 1. Brief Introduction to Deep Learning for Object Recognition Using MATLAB Mohammad Tamim Alkhodary Mar 7, 2018 1
- 2. Objectives • Get Inspired by new innovation and the transformation of the technology • Understand the concept of Deep Learning and it’s relation to machine learning and Artificial Intelligence • Get stared with Deep Neural Network 2
- 3. Contents Introduction to AI and Machine Learning What is Deep Learning How does deep learning works Applications of Deep Learning MATLAB Example 3
- 4. Artificial Intelligence (AI) 4 • AI will 'transform or destroy' society • AI could be the "worst event in the history of our civilization" unless society finds a way to control its developments
- 5. 5
- 6. 6
- 7. What is Deep Learning? • Deep learning is a Machine learning technique that learn features and tasks directly from data • Date could be 7 images, text, sounds, or data points
- 8. Example of Deep Learning 8
- 9. N to N learning 9
- 10. Why now? 10
- 11. Why now? 11
- 12. Steps for Full Deep Learning Netorwk • Objects • Model Building • Data • Different Senario Training • right away Implement 12
- 17. 17
- 18. Applications of Deep Learning 18
- 19. Applications of Deep Learning 19
- 22. Image recognition Application • Self Driving Car • Health: cancer cell prediction • Quality control and defected item recognition • Counting 22
- 23. 23
- 24. 24
- 25. 25
- 26. How to build Deep Neural Network Decide on the application Build the feature extraction layers (Deep layers) Train the network Export Network cof Set-up Stage
- 27. How to build Deep Neural Network 27 Import DNN conf. data adjust your new data according to parameters start using it Implementation Stage
- 28. Alex Net 2012 • In MATLAB Add-ins search for Pre-trained Network • Install AlexNet 28
- 30. Needed Packages for MATLAB 30 • MATLAB Support Package for USB Webcams • AlexNet pre-trained network • Statistics and Machine Learning Toolbox
- 31. Basic MATLAB commands • net = Alexnet % network ojects • Lyers = net.Layers • Input layer = Lyers(1) % first layer input layer • ClassLayer = ly(end) % is the class name • CategoryNames = ClassLayer.ClassNames • [pred, percent] = classify(net, ImagePointer) 31
- 32. Transfer Learning checklist A network Training Data Algorithm options Batch size Max iteration Learning Rate 32
- 33. Resources • MATLAB Machine learning help document • Machine learning for self-driving cars MIT-open course • Krizhevsky, Alex, Ilya Sutskever, and Geoffrey E. Hinton. "ImageNet Classification with Deep Convolutional Neural Networks." Advances in neural information processing systems. 2012. 33
- 35. How to train a network 35
- 36. 36 imds = imageDatastore('file*.jpg’) img = readimage(imds,07) imageDatastore('file*.jpg’) imageDatastore(‘foldername’,’include sub folders’, true)