基于Google云平台的TensorFlow图像理解

课程概况

This is the third course of the Advanced Machine Learning on GCP specialization. In this course,

We will take a look at different strategies for building an image classifier using convolutional neural networks. We’ll improve the model’s accuracy with augmentation, feature extraction, and fine-tuning hyperparameters while trying to avoid overfitting our data. We will also look at practical issues that arise, for example, when you don’t have enough data and how to incorporate the latest research findings into our models.

You will get hands-on practice building and optimizing your own image classification models on a variety of public datasets in the labs we’ll work on together.

Prerequisites: Basic SQL, familiarity with Python and TensorFlow

课程大纲

Welcome to Image Understanding with TensorFlow on GCP

In this introductory module you will learn about the rapid growth in high-resolution image data available and the types of applications that it can be applied to. We’ll also cover image data as inputs to your model.

Linear and DNN Models

We’ll start with a brief introduction where we’ll cover the image dataset you will be using for part of this course. Then we’ll tackle an image classification problem with a linear model in TensorFlow. After that we’ll move onto tackling the same problem using a Deep Neural Network. Lastly, we’ll close with a discussion and application of dropout which is a regularization technique for neural networks to help prevent them from memorizing our training dataset.

Convolutional Neural Networks (CNNs)

This module will introduce Convolutional Neural Networks or CNNs for short, and get you started with implementing CNNs using TensorFlow. Since 2012, CNN based systems achieved unparalleled performance on tasks like image recognition and even at playing the ancient board game of Go against the top human champions.

Dealing with Data Scarcity

In this module, we’ll focus on data scarcity, what it is, why it’s important, and, before moving onto building ML models, what you need to do about it.

Going Deeper Faster

In this module, you will learn how to train deeper, more accurate networks and do such training faster.You will learn about common problems that arise when training deeper networks, and how researchers have been able to address these issues.

Pre-built ML Models for Image Classification

Welcome to the last module of the Image Classification course. Now that you have build your own image classifiers using linear, DNN, and CNN models with TensorFlow, it’s time to experiment with pre-built image models. In most cases, you will want to try these before investing your time in developing custom TensorFlow code for a model.

Summary

In this final module, we will review the core concepts covered in this image classification course. You will recall creating classifiers with linear models, DNNs, DNNs with Dropout, Convolutional Neural Networks (CNNs), and lastly with pre-built models like the Cloud Vision API and AutoML Vision.