My learning habit is to go from abstraction to concreteness, from system to node, and systematically sort out the main points of using DNN/CNN for image classification. If there are unfamiliar terms in the figure, it is the part that needs to be supplemented.
Some books and materials were referred to in the process of making this picture, as follows:
"Core Image Recognition Technology and Case Combat in Deep Learning" – There are three words
"Mathematical Practice in Image Processing" – Zuo Fei
"Machine Learning for Intelligent Image Processing Using OpenCV and Python" --Michael Beyeler
https://neuralnetworksanddeeplearning.com/ I think the best BP principle explanation