November 2019
Intermediate to advanced
346 pages
9h 36m
English
The greatest potential for a counterfeiting solution lies in obtaining a large dataset of images and using deep learning technology. In a regime where the dataset is relatively small, as is the case here, however, feature-engineering is mandatory. We begin attacking our problem by loading and then reading in a dataset into pandas (Steps 1 and 2). In the case of this dataset, a wavelet transform tool was used to extract features from the images. Next, in Steps 3 and 4, we train-test split the data and gather it into arrays. Finally, we fit and test a basic classifier on the dataset in Steps 5 and 6. The high score (98%) suggests that the features extracted for this dataset are indeed able to distinguish between authentic ...