In this hands-on workshop, you'll use FiftyOne to run the full rare-class mining loop end-to-end on a large unlabeled image pool: compress the pool with near-duplicate detection, embed images with a modern vision backbone, mine candidate positives via seeded similarity from a tiny labeled set, confirm them through targeted human review, and prioritize the survivors for annotation using representativeness and uniqueness scores. You'll then fine-tune a detector on the prioritized examples and come back to FiftyOne to verify the missed-mode clusters actually got covered.

The punchline: a few dozen images chosen by sequencing exploit and explore through embedding space beats thousands of randomly sampled ones, every time.