Search icon CANCEL
Subscription
0
Cart icon
Your Cart (0 item)
Close icon
You have no products in your basket yet
Save more on your purchases! discount-offer-chevron-icon
Savings automatically calculated. No voucher code required.
Arrow left icon
All Products
Best Sellers
New Releases
Books
Videos
Audiobooks
Learning Hub
Newsletter Hub
Free Learning
Arrow right icon
timer SALE ENDS IN
0 Days
:
00 Hours
:
00 Minutes
:
00 Seconds
Arrow up icon
GO TO TOP
TensorFlow Deep Learning Projects

You're reading from   TensorFlow Deep Learning Projects 10 real-world projects on computer vision, machine translation, chatbots, and reinforcement learning

Arrow left icon
Product type Paperback
Published in Mar 2018
Publisher Packt
ISBN-13 9781788398060
Length 320 pages
Edition 1st Edition
Languages
Concepts
Arrow right icon
Authors (2):
Arrow left icon
 Shanmugamani Shanmugamani
Author Profile Icon Shanmugamani
Shanmugamani
Alexey Grigorev Alexey Grigorev
Author Profile Icon Alexey Grigorev
Alexey Grigorev
Arrow right icon
View More author details
Toc

Table of Contents (16) Chapters Close

Title Page
Packt Upsell
Contributors
Preface
1. Recognizing traffic signs using Convnets FREE CHAPTER 2. Annotating Images with Object Detection API 3. Caption Generation for Images 4. Building GANs for Conditional Image Creation 5. Stock Price Prediction with LSTM 6. Create and Train Machine Translation Systems 7. Train and Set up a Chatbot, Able to Discuss Like a Human 8. Detecting Duplicate Quora Questions 9. Building a TensorFlow Recommender System 10. Video Games by Reinforcement Learning 1. Other Books You May Enjoy Index

Image captioning approaches


There are several approaches to captioning images. Earlier methods used to construct a sentence based on the objects and attributes present in the image. Later, recurrent neural networks (RNN) were used to generate sentences. The most accurate method uses the attention mechanism. Let's explore these techniques and results in detail in this section. 

Conditional random field

Initially a method was tried with the conditional random field (CRF) constructing the sentence with the objects and attributes detected in the image. The steps involved in this process are shown as follows:

System flow for an example images (Source: http://www.tamaraberg.com/papers/generation_cvpr11.pdf)

CRF has limited ability to come up with sentences in a coherent manner. The quality of generated sentences is not great, as shown in the following screenshot:

The sentences shown here are too structured despite getting the objects and attributes correct.

Note

Kulkarni et al., in the paper http://www...

lock icon The rest of the chapter is locked
Register for a free Packt account to unlock a world of extra content!
A free Packt account unlocks extra newsletters, articles, discounted offers, and much more. Start advancing your knowledge today.
Unlock this book and the full library FREE for 7 days
Get unlimited access to 7000+ expert-authored eBooks and videos courses covering every tech area you can think of
Renews at £13.99/month. Cancel anytime
Visually different images