I am working on software which deblurs the motion blur created by camera movement. I've surveyed some research papers and determined this process requires deep learning and CNN. Now I'm looking for some books that would be useful in getting a more complete picture of the process.
If you are new to neural networks, I recommend studying the free online book Neural Networks and Deep Learning. It teaches you the basic concepts and the underlying math - a great starting point to dig deeper.
Once you understand the basic concepts, I recommend watching the Stanford Lecture CS231n about Convolutional Neural Networks for Visual Recognition. It consists of 16 lectures and teaches you most of the things you'll need to know for your project. The also provide all the slides - check out the links below each video.
Analyzing existing projects and their documentation is a good next step for deeper understanding of what's state of the art in CNN. You could start with this article about Q-Learning based on a CNN architecture, but there are many more out there.
More specific questions will come during your project. You'll find more information in scientific papers and blogs from researchers. Feel free to ask here as well, if you don't find what you are looking for elsewhere.