I don't think there's a single "best" book, especially not one that would at the same time suit the experts and beginners, so let's go with 3 (subjectively chosen) examples.
The most comprehensive title out there is probably "Deep Learning" by Ian Goodfellow, Yoshua Bengio and Aaron Courville - it gets you from basics of machine learning (algebra, probability, information theory), through neural networks to more complex concepts like autoencoders, generative models or Monte Carlo methods. In my opinion it's more academic than practical, containing a good deal of solid theory. Bonus: you can read it for free online here: http://www.deeplearningbook.org/
Francois Chollet's "Deep Learning with Python" gets great reviews, as it contains a ton of practical examples. Not so much theory-oriented, I would say it's a better pick for a beginner (disclaimer: I haven't read it myself). It has chapters on computer vision as well as sequences - see the table of contents here: https://www.manning.com/books/deep-learning-with-python
I could also add Kevin Murphy's "Machine Learning: A Probabilistic Perspective" - it builds the concept of machine learning in incremental steps from probability theory (which the reader should feel fairly comfortable in to begin with), Gaussian models, Bayesian statistics, up to Markov models. Final chapter is on deep learning - check the table of contents: https://mitpress.mit.edu/books/machine-learning-0