Free AWS AIF-C01 Certified AI Practitioner Actual Exam Questions - Question 8 Discussion
high accuracy and must minimize the risk of incorrect annotations.
Which solution will meet these requirements?
I’m thinking option C might not fit well because while Amazon Rekognition does image recognition, it doesn’t specifically focus on improving annotation accuracy or validation. Also, D with QuickSight seems off since it’s more about data visualization, not annotation or image processing. B talks about data augmentation, which helps expand datasets but doesn’t directly address minimizing incorrect annotations. So really, the question boils down to solutions that ensure annotation accuracy rather than just processing or expanding data. Would you agree the focus is more on quality control than on
A, because human review reduces annotation errors better than automated options.
It’s A because human-in-the-loop ensures errors get caught, unlike C which is fully automated and might miss details in protective eyewear images. The question emphasizes minimizing annotation mistakes, so automation alone isn't enough.
Option A, humans checking labels directly reduces annotation errors best here.
I get why A makes sense since humans double-checking can reduce errors. But what if the dataset is huge and time-sensitive? Then, relying on human validation might slow things down a lot. C could be tempting if you already have a solid model for recognizing eyewear features, so maybe a mix of automated recognition with some human checks could balance speed and accuracy. Does the question’s wording about “minimizing incorrect annotations” rule out fully automatic options like C?
This one feels like option A. Human-in-the-loop validation usually means more accuracy and fewer mistakes for annotations. The others don’t really fit the problem as well.