News for hackers. Stuff that matters.
Re:Advancing Research on Video Understanding with the YouTub... by AndrewKemendo on Monday February 06, 2017 @04:27PM (#13584225)
210,000 annotated videos is amazingly large, and as others asked, I really wonder how they did their training - probably offloaded it to users more than likely. Really glad to see these data sets going out to the world.

It is a huge problem that Google/Apple/Facebook/Amazon have the bulk of the data gathering ability, which just means they have 100x the ability to iterate machine learning capabilities and applications. For them it's a huge win and exactly what I would be doing if I were them, but it leaves startups and smaller machine learning companies like us out there scrambling for new data sources that aren't touched by those (hint: there basically are none).

edit: This data also really helps us start teaching our machine vision systems mechanical and physical processes, as described here: https://arxiv.org/abs/1603.01312

Re:Advancing Research on Video Understanding with the YouTub... by pmalynin on Monday February 06, 2017 @07:09PM (#13585683)
The official site ( https://research.google.com/youtube-bb/explore.html ), has interesting examples.

One of them is a guy drinking what appears to be urine from a toilet with a spoon.

Re:Advancing Research on Video Understanding with the YouTub... by teLeopardthy on Monday February 06, 2017 @08:16PM (#13586069)
It seems odd that only one object of each class is tracked in a video segment. I'd imagine that it would be a limitation for algorithms that generate bounding boxes for each class, as they might be penalized for correctly finding the non-tracked instances of a given class. Is only tracking a single instance of a class standard for this kind of dataset?
Re:Advancing Research on Video Understanding with the YouTub... by rasz_pl on Monday February 06, 2017 @09:31PM (#13586469)
All this amazing automation and deep learning and Google still spams me with pewdiepie and other meme lord YT drama garbage suggestions despite clicking "not interested" hundreds of times.
Re:Advancing Research on Video Understanding with the YouTub... by jalk on Monday February 06, 2017 @02:59PM (#13583283)
Wondering who and how they manually boxed/labeled the objects. Sounds like an amazing job :)
Re:Advancing Research on Video Understanding with the YouTub... by Guyag on Monday February 06, 2017 @08:27PM (#13586131)
Their method is detailed in the whitepaper [1], mainly section 3.2. TL;DR mturk for high level annotations (frame contains/does not contain object in class), and internal annotators for bounding boxes.

[1] https://arxiv.org/pdf/1702.00824.pdf

Re:Advancing Research on Video Understanding with the YouTub... by 0x00000000 on Monday February 06, 2017 @03:26PM (#13583509)
Probably done with the new type of captcha (select all boxes with ____)
Re:Advancing Research on Video Understanding with the YouTub... by maxmcd on Monday February 06, 2017 @03:49PM (#13583743)
The bounding boxes look more exact than that. I have also only seen that new type of captcha with images that look like they are from streetview.