Image Classification with TensorFlow and Keras
Published by Pearson
Train a Clothes Classification Model and Deploy It on the Cloud
Deep learning achieves the best performance for computer vision, natural language processing, and recommendation tasks.
In this workshop, you'll learn how to train a deep learning model for image classification. We will develop a model for categorizing pictures of clothes into 10 different classes: from T-shirts to jackets. After finishing, you’ll be able to train a similar model.
What you’ll learn and how you can apply it
By the end of the live online course, you’ll understand:
- What is a convolutional network and how it can be used for image classification
- What is transfer learning and how it can help us create good neural network model with little data
- What is dropout and data augmentation and how they help create better models
And you’ll be able to:
- Train a neural network model for image classification using TensorFlow and Keras
- Use neural networks created by others and adapt them for your use cases
- Tune the parameters of a model to achieve the best quality of the model
- Use Keras to train a model for classifying images of clothes (T-shirts, shoes, outerwear, and others)
- Fine-tune a pretrained model
- Adjust it to achieve the best performance
- Generate more image examples from the existing dataset with data augmentation
This live event is for you because...
- You’re a data scientist or a software engineer interested in machine learning
- You want to learn more about using deep learning for image classification
Prerequisites
- You should be comfortable following code in Python.
- Basic knowledge of Python is required (for loops, data structures like lists and dictionaries, imports).
- Experience with Python tools and libraries like Jupyter, NumPy and Matplotlib is helpful, but not required.
- We use Docker in one section. Knowledge of Docker is helpful but not required.
Course Set-up
You can just watch the live coding session. To reproduce it, you will need:
- Jupyter Notebook. The simplest way to get it is to install Anaconda.
- TensorFlow. Alexey will show how to install it.
- The code used in this course is available on https://github.com/alexeygrigorev/deep-learning-keras-aws-lambda
Recommended Preparation
- Read: Deep Learning Illustrated [Book] by Jon Krohn
Recommended Follow-up
- Watch: Deep Learning with TensorFlow, Keras, and PyTorch (Video) by Jon Krohn
Schedule
The time frames are only estimates and may vary according to how the class is progressing.
Segment 1: Introduction (15 minutes)
- About the instructor
- What we’ll cover today
- How to set up the environment — AWS, Anaconda + Jupyter, TensorFlow + Keras,
- Alternative setup — Google Collab
- The first look at the dataset
Segment 2: Training the first model (60 minutes)
- Loading and preparing data
- Transfer learning with Keras
- Training a small model on a GPU
- Saving the model and using it
Break (10 minutes)
Q&A (10 minutes)
Segment 3: Improving the model (60 minutes)
- Adding more layers
- Checkpointing: saving the best version of our model
- Making model generalize better: adding dropout
- Generating more data with data augmentation
- Training a larger model
- Testing the model: applying it to test data
Break (5 minutes)
Q&A (10 minutes)
Wrapping up, next steps (10 minutes)
Your Instructor
Alexey Grigorev
Alexey Grigorev works as a principal data scientist at OLX. He runs DataTalks.Club — a community for data enthusiasts. Alexey has written books about machine learning including his most recent, Machine Learning Bookcamp, which is a book for software engineers who want to get into machine learning. He lives in Berlin with his wife and son.