Walk through a Notebook for computer vision

Loading...

Aus dem Kurs von deeplearning.ai

Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning

1135 Bewertungen

deeplearning.ai

Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning

1135 Bewertungen

If you are a software developer who wants to build scalable AI-powered algorithms, you need to understand how to use the tools to build them. This course is part of the upcoming Machine Learning in Tensorflow Specialization and will teach you best practices for using TensorFlow, a popular open-source framework for machine learning. The Machine Learning course and Deep Learning Specialization from Andrew Ng teach the most important and foundational principles of Machine Learning and Deep Learning. This new deeplearning.ai TensorFlow Specialization teaches you how to use TensorFlow to implement those principles so that you can start building and applying scalable models to real-world problems. To develop a deeper understanding of how neural networks work, we recommend that you take the Deep Learning Specialization. The full deeplearning.ai TensorFlow Specialization will be available later this year, but you can get started with Course 1, Introduction to Tensorflow for AI, ML and DL, available now on Coursera.

Aus der Unterrichtseinheit

Introduction to Computer Vision

Welcome to week 2 of the course! In week 1 you learned all about how Machine Learning and Deep Learning is a new programming paradigm. This week you’re going to take that to the next level by beginning to solve problems of computer vision with just a few lines of code! Check out this conversation between Laurence and Andrew where they discuss it and introduce you to Computer Vision!

A Conversation with Andrew Ng2:13

An Introduction to computer vision2:02

Writing code to load training data2:18

Coding a Computer Vision Neural Network2:25

Walk through a Notebook for computer vision3:04

Using Callbacks to control training1:41

Walk through a notebook with Callbacks1:13

Treffen Sie die Kursleiter

Laurence Moroney
AI Advocate
Google Brain

Okay. So you just saw

how to create a neural network that gives

basic computer vision capabilities

to recognize different items of clothing.

Let's now work through a workbook

that has all of the code to do that.

You'll then go through this workbook yourself

and if you want you can try some exercises.

Let's start by importing TensorFlow.

I'm going to get the fashion MNIST data

using tf.kares.datasets.

By calling the load data method,

I get training data and labels

as well as test data and labels.

For more details on these,

check back to the previous video.

The data for a particular image is a grid of values from

zero to 255 with pixel Grayscale values.

Using matplotlib, I can plot

these as an image to make it easier to inspect.

I can also print out the raw values

so we can see what they look like.

Here you can see the raw values for

the pixel numbers from zero to 255,

and here you can see the actual image.

That was for the first image in the array.

Let's take a look at the image at index 42 instead,

and we can see the different pixel values

and the actual graphic.

Our image has values from zero to 255,

but neural networks work better with normalized data.

So, let's change it to between zero and one

simply by dividing every value by 255.

In Python, you can actually divide

an entire array with one line of code like this.

So now we design our model.

As explained earlier, there's

an input layer in the shape of

the data and an output layer in the shape of the classes,

and one hidden layer that

tries to figure out the roles between them.

Now we compile the model to finding

the loss function and the optimizer,

and the goal of these is as before,

to make a guess as to what the relationship

is between the input data and the output data,

measure how well or how badly it

did using the loss function,

use the optimizer to generate a new gas and repeat.

We can then try to fit

the training images to the training labels.

We'll just do it for five epochs to be quick.

We spend about 25 seconds training it over

five epochs and we end up with a loss of about 0.29.

That means it's pretty accurate in guessing

the relationship between the images and their labels.

That's not great, but considering it was done in

just 25 seconds with

a very basic neural network, it's not bad either.

But a better measure of performance can

be seen by trying the test data.

These are images that the network has not yet seen.

You would expect performance to be worse,

but if it's much worse, you have a problem.

As you can see, it's about 0.345 loss,

meaning it's a little bit less accurate on the test set.

It's not great either, but we

know we're doing something right.

Your job now is to go through the workbook,

try the exercises and see by tweaking

the parameters on the neural network

or changing the epochs,

if there's a way for you to get it above 0.71 loss

accuracy on training data and 0.66 accuracy on test data,

give it a try for yourself.

Durchsuchen Sie unseren Katalog

Melden Sie sich kostenlos an und erhalten Sie individuelle Empfehlungen, Aktualisierungen und Angebote.

Coursera arbeitet mit erstklassigen Universitäten und Organisationen zusammen, um Online-Kurse anzubieten und dadurch universellen Zugriff auf die weltweit beste Ausbildung zu ermöglichen.

© 2019 Coursera Inc. Alle Rechte vorbehalten.

Aus dem App Store herunterladen

Erhältlich bei Google Play