I tried a Kaggle machine learning contest ("Hotel-ID to Combat Human Trafficking 2022"), and learned some things about Kaggle and modern image recognition technology, including: dominance of CNNs, image augmentation, picking a model (timm, efficientnet, arcface nearest-neighbor classification), and more.
I'll walk through what Kaggle is, why participate in a Kaggle contest, and a bit about what it's like (including an example notebook). You don't need much AI background, but we will look at some code. If I have time, I'll also present some learnings about image recognition and Kaggle.
I got as high as #7 on the leaderboard, and finished at #9 (of 83) with an accuracy of 0.554 (of 1.0).
There may be people in the room who know more about Kaggle or image classification than me. Time permitting, we can chat and share knowledge.
I have worked at twin cities startups (Net Perceptions, Orasi Medical, code42, Blue Shift Labs), at larger companies (Google, Amazon, Pinterest), and as a fellow in the GroupLens research lab at the University of Minnesota. I was a data scientist at Pinterest for six years (from startup to big). See more at linkedin.