Hacker News Reader

Comment by surfacenexus | Hacker News Reader

You even caught the bottles reflected in the sunglasses, incredible!

Totally; the model does a good job finding what you teach it to find!

On the flip side, someone in the YouTube comments mentioned that we missed the ones on the wheels and license plates. So I added them to the dataset this morning and kicked off another train job.

The AWS bill is $2 milion

I also noticed some false positives. It is a good demonstration of the difficulty of the task: there is no easy, automated, out of the box solution that will arrive at the correct answer. You could make a decent approximation if you could correlate frames and track individual bottles but even that is beyond the scope of this demo.

Not to mention the 90% recall, meaning that the model is missing 10% of the bottles in the annotated data.

Cool blog post, cool tutorial, and great tool but its not going to win any contest where you need to guess an exact number.

Are those two seperate bottles being reflected once, or one bottle being reflected twice?

Good question. And do they count as separate bottles if the un-reflected version is shown in a different frame?

The official rules state that each reflected bottle is its OWN "counted bottle".

I was otherwise rather confused by the "count each bottle only once": there are a few in marquees and such where it's hard to tell if it's a repeat or a new bottle, frankly.

Reply on news.ycombinator.com