It is said that for a machine learning model to be successful, you need to have good data. While this is true (and quite obvious), it is extremely difficult to define, build, and maintain good data. Let me share with you the unique processes I have learned over several years building an ever-growing image classification system and how you can apply these techniques to your own application.
With persistence and diligence, you can avoid the classic “garbage in, garbage out,” maximize the accuracy of your model, and demonstrate real business value.
In this series of articles, I will delve into the care and feeding of a single-label, multi-class image classification application and what it takes to achieve the highest level of performance. I won't go into coding or specific user interfaces, just the core concepts you can incorporate to meet your needs with the tools at your disposal.
Below is a brief description of the items. You will notice that the model is the last one in the list, since, first of all, we must focus on selecting the data:
- Part 1: The data: labeling standards, classes and subclasses
