While I was doing an object detection project, I have encountered the problem of getting FALSE POSITIVES and FALSE NEGATIVES. After days of research on StackOverflow, I figured out that I need to collect more negative images or background images.I decided to document this process so other people could easily solve this issue and the result of documentation is this. After training the model with Negative/Background images, my FP/FN rates were normalized so that in video frames I started getting fewer FPs. All of us, machine learning developers get experience by getting hands dirty - this is clear to all of us. But I haven't seen(probably missed) any video tutorials or examples on books showing how to collect background images and why we need them at all.
So here is the question: Okay, so every experienced ML engineer knows what is the FP/FNs are, and their prevention methods. But why this topic is less known and taught within popular object detection tutorials and books? Or am I missing something?