TensorFlow can do some amazing things when it comes to computer vision.
We just published a full course on the freeCodeCamp.org YouTube channel that will teach you how to use TensorFlow 2 for computer vision applications.
Nour Islam Mokhtari created this course. Nour is a Machine Learning Engineer and experienced teacher.
The course shows you how to create two computer vision projects. The first involves an image classification model with a prepared dataset. The second is a more real-world problem where you will have to clean and prepare a dataset before using it.
Here are the topics covered in this course:
- Why learn Tensorflow
- We will be using an IDE and not notebooks
- Visual Studio Code (how to download and install it)
- Miniconda - how to install it
- Miniconda - why we need it
- How are we going to use conda virtual environments in VS Code?
- Installing Tensorflow 2 (CPU version)
- Installing Tensorflow 2 (GPU version)
- What do we want to achieve?
- Exploring MNIST dataset
- Tensorflow layers
- Building a neural network the sequential way
- Compiling the model and fitting the data
- Building a neural network the functional way
- Building a neural network the Model Class way
- Things we should add
- Restructuring our code for better readability
- First part summary
- What we want to achieve
- Downloading and exploring the dataset
- Preparing train and validation sets
- Preparing the test set
- Building a neural network the functional way
- Creating data generators
- Instantiating the generators
- Compiling the model and fitting the data
- Adding callbacks
- Evaluating the model
- Potential improvements
- Running prediction on single images
Watch the full course below or on the freeCodeCamp.org YouTube channel (4.5-hour watch).