Implementations in TensorFlow and Keras

Thanks to the efforts that the authors put into creating a clear architecture, VGG-16 and VGG-19 are among the simplest classifiers to reimplement. Example code can be found in the GitHub folder for this chapter, for educational purposes. However, in computer vision, as in many domains, it is always preferable not to reinvent the wheel and to instead reuse existing tools that are available. The following paragraphs present different preimplemented VGG solutions that you can directly adapt and reuse.