On this last a part of my series, I’ll share what I even have learned on choosing a model for image classification and easy methods to nice tune that model. I may even show how you’ll be able to leverage the model to speed up your labelling process, and eventually easy methods to justify your efforts by generating usage and performance statistics.
In Part 1, I discussed the technique of labelling your image data that you just use in your image classification project. I showed how define “good” images and create sub-classes. In Part 2, I went over various data sets, beyond the standard train-validation-test sets, with benchmark sets, plus easy methods to handle synthetic data and duplicate images. In Part 3, I explained easy methods to apply different evaluation criteria to a trained model versus a deployed model, and using benchmarks to find out when to deploy a model.
Model selection
To date I even have focused plenty of time on labelling and curating the set of images, and likewise evaluating model performance, which is like putting the cart before the horse. I’m not trying to attenuate what it takes to design a large neural network — that is an important a part of the applying you’re constructing. In my…
