What you'll learn:
- Use pre-trained TensorFlow models to solve Computer Vision and Natural Language Processing problems
- Classify images of flowers using Convolutional Neural Networks
- Detect over 80 different objects in images
- Apply style transfer to images
- Build a GAN to complete the missing parts of images
- Recognize actions in videos
- Classify sentiments in texts
- Use information retrieval techniques to return similar documents
- Classify over 500 audio events
Deep Learning is the application of artificial neural networks to solve complex problems and commercial problems. There are several practical applications that have already been built using these techniques, such as: self-driving cars, development of new medicines, diagnosis of diseases, automatic generation of news, facial recognition, product recommendation, forecast of stock prices, and many others! The technique used to solve these problems is artificial neural networks, which aims to simulate how the human brain works. They are considered to be the most advanced techniques in the Machine Learning area.
One of the most used libraries to implement this type of application is Google TensorFlow, which supports advanced architectures of artificial neural networks. There is also a repository called TensorFlow Hub which contains pre-trained neural networks for solving many kinds of problems, mainly in the area of Computer Vision and Natural Language Processing. The advantage is that you do not need to train a neural network from scratch! Google itself provides hundreds of ready-to-use models, so you just need to load and use them in your own projects. Another advantage is that few lines of code are needed to get the results!
In this course you will have a practical overview of some of the main TensorFlow Hub models that can be applied to the development of Deep Learning projects! At the end, you will have all the necessary tools to use TensorFlow Hub to build complex solutions that can be applied to business problems. See below the projects that you are going to implement:
Classification of five species of flowers
Detection of over 80 different objects
Creating new images using style transfer
Use of GAN (generative adversarial network) to complete missing parts of images
Recognition of actions in videos
Text polarity classification (positive and negative)
Use of a question and answer (Q&A) dataset to find similar document
Audio classification
All implementations will be done step by step using Google Colab online, so you do not need to worry about installing and configuring the tools on your own machine! There are more than 50 classes and more than 7 hours of videos!
