Practical Deep Learning for Cloud, Mobile, and Edge: Real-World AI & Computer-Vision Projects Using Python, Keras & TensorFlow

by Anirudh Koul, Siddha Ganju, Meher Kasam

** Featured as a learning resource on the official Keras website **

Whether you're a software engineer aspiring to enter the world of deep learning, a veteran data scientist, or a hobbyist with a simple dream of making the next viral AI app, you might have wondered where to begin. This step-by-step guide teaches you how to build practical deep learning applications for the cloud, mobile, browsers, and edge devices using a hands-on approach. If your goal is to build something creative, useful, scalable, or just plain cool, this book is for you.

Relying on decades of combined industry experience transforming deep learning research into award-winning applications, Anirudh Koul, Siddha Ganju, and Meher Kasam guide you through the process of converting an idea into something that people in the real world can use.
List of Chapters Guest-contributed Content The book features chapters from the following industry The book also features content contributed by several industry veterans including François Chollet ( Keras , Google ), Jeremy Howard ( Fast.ai ), Pete Warden ( TensorFlow Mobile ), Anima Anandkumar ( NVIDIA ), Chris Anderson ( 3D Robotics ), Shanqing Cai ( TensorFlow.js ), Daniel Smilkov ( TensorFlow.js ), Cristobal Valenzuela ( ml5.js ), Daniel Shiffman ( ml5.js ), Hart Woolery ( CV 2020 ), Dan Abdinoor ( Fritz ), Chitoku Yato ( NVIDIA Jetson Nano), John Welsh ( NVIDIA Jetson Nano), and Danny Atsmon ( Cognata ).

Genres: Technical, Artificial Intelligence, Programming, Computer Science, Technology, Computers, Coding

583 pages, Paperback

Published November 26, 2019

Reviews

[Krishnan Ajay · Rating 5 out of 5]

TLDR; Best book to understand and use Deep Learning in the industry. This book provides a bridge between what’s taught in courses and what is actually done in the industry.
I was just finishing my Junior year with a lot of theoretical knowledge of Deep Learning but not enough practical experience to showcase this during interviews. Most resources on the internet only gave a theoretical understanding through equations, whereas this book has been completely practical and goal-oriented since its first page. This book does an amazing job at telling you what exactly you need in order to deploy your model, scale it further and analyze its performance as it grows.

Each chapter becomes a separate project by itself on different areas in Deep Learning. The authors also made this book extremely beginner-friendly and easy to understand such that implementing and testing every concept was not half as complex as it was in theory. There are cheatsheets present at every chapter providing tips and techniques for every situation an ML Engineer building AI for an edge device might find themselves in. After an in-depth explanation of a certain process/concept in every chapter, there is also a Case Studies section about the same concept being deployed in a popular Silicon Valley startup / a tech giant with an explanation about how this concept works in the company’s use-case.

My favorite chapters were those on building our own Self Driving car and Building an Autonomous Car using Reinforcement Learning with AWS DeepRacer. These chapters gave me clear insight into how AI works in the field of Self Driving Vehicles and also pointed me towards implementing them.

The book teaches you how to work with all the state-of-the-art libraries, and tools such as TensorFlow, Keras, TFLite for Android, CoreML for iPhone, and more, suited for edge devices.
The information on making full use of edge devices up giving me a huge boost in interviews and subsequent interviews too. 10/10 would recommend if you’re looking to apply all the theory you’ve learned from courses into real life.

[Paco Nathan · Rating 5 out of 5]

Excellent into to deep learning, based on open source tools and practical use cases. Loved the book so much that I got to be a technical reviewer :)

[Satyarth Praveen · Rating 5 out of 5]

Have some idea regarding deep learning but not enough experience with the projects and deployment? This is the book you NEED!

This is a brilliant book to gain hands-on experience with some deep learning projects and answer any of your deployment related questions - this includes scalability, efficiency, and beautiful visualizations. I have bought this book for my sister and recommend it to anyone willing to get started with AI projects. I have been using code snippets from this book for a lot of my deep-learning projects. Plus plus plus, the book is easy to read and engages the reader with random puns!

[Rudy Venguswamy · Rating 5 out of 5]

This book provide a deeply informative and influential toolkit for teams and individual practitioners of machine learning and artificial intelligence to update their skill set to the more modern standards of AI in industry. It condenses the subject matter expertise of the authors ranging including training optimal models, item similarity, model deployment, and computer vision interpretability into easily understood concepts coupled with code that I was able to apply and tweak for my own projects. As a recent graduate in the field of machine learning, understanding how I could apply computer vision to my projects, both in school and in work, was quite valuable to me and provides an organized framework through which I could think about how to design and tackle a computer vision problem using ML. In my circle in particular, this book has become a go-to guide for computer vision and ML development.

Our team was able to get up to speed and build a reverse image search pipeline for NASA and through the examples provided on efficient code design, were able to upgrade our tool to scale to petabyte-levels of information. Unlike the purely theoretical literature that is available in abundance on computer vision, this practical guide contains the authors' insights on bringing these theoretical concepts to industry-readiness and, at least in my team, has had a disproportionate influence on our machine learning strategy and deployment.

[Fanyue Xia · Rating 5 out of 5]

Being a first-year undergraduate student without much prior knowledge of artificial intelligence, this book really brought my confidence up.

Different from other books and courses about AI I had studied before, this book focuses on practical applications with the help of Python, TensorFlow and Keras ecosystem. Almost every chapter starts with an interesting real-life scenario and problems needed to be solved. It then includes step by step instructions on how to approach the problem with complementary codes and explanation as well as graphical demonstrations of various APIs. For me, it is a rewarding experience to take advantage of numerous resources recommended by the book, such as experimenting with the codes and data myself on Google Colab, exploring other related projects being developed on Github, and training my own model using Google Cloud Vision. Moreover, case studies at the end of each chapter, which elaborates on how the methods presented are being employed by big companies to develop important real-life applications, sufficiently establish the power and availability of machine learning, making me determined to explore machine learning further as my additional major and potential career option. 

Overall, this book is a fantastic guide for readers to understand the concept of machine learning, to equip themselves with the ability to really contribute to the machine learning community, and to keep up with its leading-edge research and application.

[Tina · Rating 5 out of 5]

I am not a computer science engineer and did not know anything about ML. However, after reading this book, I can run and train my own customized models to solve problems related to my field (management). The book is much fun to read and the examples used help the user connect things to the real world and not just computer geekery. I would recommend this book to anyone irrespective of their experience with Machine Learning. It is interesting, informative, insightful, and encouraging for anyone to tackle this interesting field.

[Nishant · Rating 5 out of 5]

Awesome Content!
I have just read couple of chapters till now, the way they have explained the ML concepts is just superb any newbie would be able to understand and relate to.

[Anurag Katakkar · Rating 5 out of 5]

TLDR: Overall, as someone that was neither a beginner, nor an expert in the field, I found this book a great read and found the content personally useful to several projects that I worked on. With a wide range of topics covered, I think this book will be useful to readers at both the theoretical, and the practical ends of the deep learning spectrum, and offers a wealth of information to beginners and experts alike.
Full review: Several Deep Learning texts exist that explain the math and theory behind the subject. But modern deep learning pipeline is much more than just data and algorithms. It includes data collection and cleaning, building models that work, and deploying them in a manner that actually make them useful to the end user, with unique challenges in each stage of the pipeline. This book focuses on the latter - while containing sufficient theory on several topics relevant to modern deep learning such as Representation Learning and Explainability, it provides a big picture perspective of how these concepts come in play while building an AI application. Even better, it provides concrete instructions and code examples in Python (in two popular software frameworks, Keras and TensorFlow) on training, testing and deploying deep learning models from scratch. Chapters 5, 6, and 7 are what truly set this book apart from other texts as they provide a lot of information about the practical aspects of being a deep learning practitioner (such as parallelizing CPU processing, caching data, and using automatic mixed precision to speed up training and inference) that are rarely spoken about outside of vague Medium instructional blogs, and Stack Overflow threads. Chapter 9, which talks about scaling an AI application to support a large number of users using various Cloud Computing platforms like AWS, and Google Cloud was especially enlightening to me - it contains information that one can otherwise only obtain through years of personal experience working in the industry.

[Dhiraj Srivastava · Rating 5 out of 5]

I am reviewing this book after reading the first 14 chapters out of a total of 17 chapters. I had a good experience with ML and DL through online websites such as coursera, udemy and my undergrad projects. But I never worked to bring my deep learning model from jupyter notebook to real-world application. I am currently pursuing my Master's in AI in agriculture science at Virginia Tech USA. As a part of my research, I have to build real-world deep learning applications. I was looking for a book which can build my foundation to deploy the models for a real-world scenario. This book bridged the gap between my theoretical knowledge and practical use. This book even helped me to write a research proposal and this research proposal got accepted by my university committee for financial support. I feel whether you are a beginner, average or experienced, you must go through this book if you are looking for a book which can help you to learn the techniques for building robust and efficient Ai apps. This book shows you the path to improve speed and accuracy as well as scaling to millions of users. With 30+ case studies and industry examples, you will become confident in solving any kind of industry problem. This book also focuses on Responsible AI. Some of the chapters that helped me are : Real time object classification on iOS, developing android apps with tensorflow lite, maximizing speed and accuracy of tensorflow ( a handy checklist). I highly recommend you to read this book if you are looking to solve and deploy real-world AI problems.

[Aaron · Rating 5 out of 5]

Exceptional book for learning the fundamentals of AI and how to program it yourself!

I’m a geography student with great interest in Artificial Intelligence and remote sensing. Three months ago, I was only familiar with high level machine learning terms, but I did not know much about Deep Learning.

This book changed that, teaching me the theoretical foundation of AI and Deep Learning and enabled me to write my own applications with Python using Tensorflow and Keras within weeks! Especially the git repository for this book is extremely useful, combining the practical part with the theoretical explanations in the book from the beginning. Post that, I was able to finish 8 weeks of Coursera TensorFlow Specialization (2 courses) in 8 days, because I was already familiar with it.

With the knowledge that I have gained working through this book, I got accepted as an AI researcher at SpaceML, detecting anomalies in remote sensing data from videos of earth in order to create alerts using Deep Learning (using Convolutional LSTM autoencoders).

Summarizing, I highly recommend this book to anyone who wants to quickly learn the fundamentals of Deep Learning and start programming it today!

[Hada]

Great guidebook for beginners to realize practical applications of deep learning.


I am a public officer supporting small and medium-sized enterprises in Japan for digital transformation, and used this book as increase my breadth of full AI lifecycle knowledge. There are many deep learning books that explain theories and looks at academic papers, but that’s so far from needs of practical use right now, especially for beginners.


If you would like to visually see each step in training and deployment and get useful information about practical application of deep learning, this book is one of the best to read. For example, you will be able to find out photos and graphics comparing which framework to use, how to scale up representative power of embeddings etc which help you practical deep learning use cases to understand easily.


Furthermore, you will be able to access and use free or reasonable services provided by IT giants like Colab, Github repo, which induces your interest and increase motivation.


This book must contribute prevalence of deep learning and the digital transformation which is necessary for the near future society.

[Shaveen Kumar · Rating 5 out of 5]

I have read several books on deep learning but I found this one particularly fun to read. Right from the beginning the authors keep the content interesting with relatable examples and as the book progresses, there is a very natural progression in difficulty while still remaining very understandable. For beginners to deep learning, I would definitely recommend this as one of the best choices to start with because the examples provided are very interesting and hands on and are explained in a manner that is very simple to understand. I also really liked how the authors focused on visualizing what the network sees right from the beginning of the book. I feel this is very important and makes the reader more invested in learning about how the network actually works rather that considering it a black box that magically produces results. I am personally a PyTorch user, but that didn't stop me from going through and running the examples. There is a well maintained github repository as well. I highly recommend this book.

[Zoe · Rating 5 out of 5]

I have a computer science background but I am a newbie to Deep Learning. I was a little bit hesitant to read this book because a book title with the word 'practical' sometimes means that you should already have somewhat background knowledge beforehand.
Well, the book wasn't super easy for me but it was definitely more approachable than I expected. I really enjoyed the sense of humor in this book. It made my reading time more relaxing - just like I got a fun and friendly next door professional as my tutor.
I sometimes got lost when reading detailed explanations, but for the most times the easy examples and clear summaries in this book saved me from confusion :) Also I loved the fact that I can easily access the codes used in the book and run them without an expensive computer setting.
I actually got to know that the Korean version of this book is now in a library in my neighborhood, so I'm gonna try it out for reviewing too.

[Prashanth Mangipudi · Rating 5 out of 5]

Firstly, really thankful for the contents of this book for my work projects. While I was wondering how to optimize the deep learning model computation, I've come across a TensorFlow User Group Summit keynote session by the author in which he explained techniques to improve training and inference. Based on that, I got to see the book which contains details in more depth. I'd say this book is a weapon for anyone who's looking to build deep learning models for edge devices or planning to deploy their model on the cloud with less latency. Also, the authors were kind enough to respond to queries and guidance on the path to getting better at deep learning. I should also mention the image search optimization techniques that are said in this book are pretty useful.

If you are the one who loves to scale your deep learning model while not losing much accuracy or generalization capabilities, this is the book for you.

[Vishwanath · Rating 4 out of 5]

Plenty of examples and links for more research. The material is too vast enough to make an all encompassing book but this delivers in terms of practical tips. Its evident the authors are practitioners with over 50+ practical tips provided as promised in the description that will find a place in any serious ML engineer repertoire. The consolidated list of tips are worth the book alone. Excellent comparisons of Raspberry Pi, Jetson Nano, and Google Coral. The reinforcement learning sections could have used some more practical examples in areas like q-learning but overall great read and reference material.

[Daron · Rating 5 out of 5]

I purchased Practical Deep Learning in the hopes that I could build my own machine learning model with a complex dataset. I had very little hands-on experience, and this incredible book gave me a firm foundation AND practical tools and challenges to learn the basics. I especially appreciated the Github repository which allowed me to run the same code that's in the book. In the end, I built a rudimentary model and presented it at work as a proof of concept. I found this book so useful that I passed along a copy to my brother. Strongly recommended as an accessible and enjoyable read, even for beginners!

[Sterling · Rating 5 out of 5]

A rather good and approachable introduction to AI/ML and deep learning.

Once of the major facets that I appreciate about this book is the fact that everything is stored in GitHub and has a matching Google Colab notebook (free GPU!), making it easy to follow along and learn by doing.

All in all, the content has been quite useful, rich in application, easy to follow along with (no excuses since Google Colab doesn't use local resources), and I've accomplished the goal I've set out with — highly recommend!

[Ladi O · Rating 5 out of 5]

There are a lot of good machine learning books for machine learning theory, however I haven't found many good ones for the practical day to day aspects for building deep learning production models. What I liked most about this book is that it indeed is very practical. This book does a good job of providing actionable advice for ML engineers/data scientist. Chapters 6 and 7, for instance, are immediately helpful especially if you are relatively new to the tensorflow/keras framework. This should definitely be in your data science library. Highly recommend

[Arman Sidhu · Rating 5 out of 5]

Being a CS major, I always wanted to get myself familiar with Machine Learning and AI. But I was confused about what resource would be the best to get started. The "learn by doing projects" approach of this book never made me feel like I was just reading a book. I recently won a hackathon by using the practical knowledge I gained from this book.

[Anuj Sharma]

I am currently reading this book, and I must say Its a pretty good book for beginners like me. Things are covered quite interestingly in it. Love the fact that its just not theory, you also have code available for hands-on experience. I am thoroughly enjoying reading it and experimenting with available code.

[Geetika Sharma · Rating 5 out of 5]

The book is definitely one of the best resources that gives insights about applied Deep Learning that too using open-source tools and plenty of examples. Would definitely recommend this to anyone looking to use Deep Learning in the industry.

[Rohit Taneja · Rating 5 out of 5]

really well structured book which helps understand not only fundamentals, but also deep-dives into the real use-cases, which help connects the theory to how DL is applied for practical tasks. Highly recommended.

[Ben · Rating 4 out of 5]

Longest Book Title Ever

This was an excellent book on some very big and key artificial intelligence players and how to use them.

I got a lot out of this big book, and look forward to furthering in my development of AI and machine learning.

3.9/5

[Pranav Kant · Rating 5 out of 5]

I went through this book as part of my ML course in university and found it very insightful. The practical approach in this book was a nice complement to the theory I was studying in class.

[Charanjeet · Rating 5 out of 5]

Great book for getting you started in practical deep learning.
Neat explaination and very easy to understand.
Highly recommended to all deep learning enthusiast.

[Martin Rowat · Rating 5 out of 5]

Chapters:
1. Exploring the landscape of articial intelligence.
2. What’s in the picture: image classification with Keras.
3. Cats versus dogs: transferlearning in 30 lines with Keras.
4. Building a reverse image search engine: understanding embeddings.
5. From novice to master predictor: maximising convolutional neural network accuracy.
6. Maximising speed and performance of TensoFlow: a handy checklist.
7. Practical tools, tips and tricks.
8. Cloud APIs for computer vision: up and running in 15 minutes.
9. Scalable inference serving on cloud with TensorFlow serving and KubeFlow.
10. AI in the browser with TensorFlow.js and ml5.js
11. Real-time object classification on iOS with CoreML and CreateML.
13. Shazam for food: developing Android apps with TensorFlow lite and MLKit.
14. Building the purrfect cat locator app eith TensoFlow object detection API.
15. Becoming a maker: exploring embedded AI at the edge.
16. Simulating a self-driving car using end-to-end deep learning with Keras.
17. Building an autonomous car in under an hour: reinforcement learning with AWS DeepRacer.

Appendix:
A crash course in convolutional neural networks.

This is a large book covering a wide range of both well developed and emergent forms of machine learning approaches to a wide range of applications. Its a big ask tackled well by the three authors and various additional contributors, all of whom are vastly experienced and respected in their fields of interest. The book offers a fascinating and engaging read for those who are seeking rather more than a basic introduction to machine learning and applied data science whether from the point of view of a lay person, or like the reviewer, someone wishes to gain more than an appreciation of the field – an exposure to the actual nuts and bolts. This book succeeds from both approaches. For the first a wide ranging account of existing and emerging techniques and tools available for various deep learning frameworks. For the second a closely integrated set of Jupyter Notebooks is made freely available. These notebooks, offer anyone with access to a web browser, the ability to carry out many of the examples written in Python with the libraries of the deep learning frameworks – such as TensorFlow and Keras. In a simple and straightforward way readers are enabled to carry out their own experimentation with the data made accessible by the authors. In other words the reader is able to DO machine learning, not just read about it!

The authors have provided a good service to the public. The book is an authoritative introduction to the constituent fields of machine learning and provides an online experience for a number of recognisable groups: the software developer who might seek to develop skills applicable in the field of AI; the data scientist to enrich their skill set and deepen knowledge of the field in order to build real projects; the student to assist in the aspiration to a career in AI through developing a portfolio of interesting projects and to help unleash creativity; the teacher since this book can supplement course work with fun real-world projects. Each of the projects presented in this book can make for great collaborative or individual work in the classroom; the robotics enthusiast for here is a survey of actual projects using emerging hardware such as Raspberry Pi, NVIDIA Jetson Nano, Google Coral and others.