We covered a whole lot of theory in the first two parts of the book. Having built a strong foundation of concepts and techniques, we started the use case-driven journey in this chapter. This chapter is the first in a series of upcoming chapters to showcase actual use cases of transfer learning in different scenarios and domains. In this chapter, we applied transfer learning to the domain of visual object identification, or, as it is popularly termed, image classification.
We started off with a quick refresher around CNNs and how the whole stage of computer-aided object identification changed once and for all with the arrival of deep learning models in 2012. We briefly touched upon various state-of-the-art image classification models, which have surpassed human performance. We also had a quick look into different benchmarking datasets utilized by academic and industry experts...
