Introducing the Open Images Dataset

Lawyers are funny:<p>&gt;Today, we introduce Open Images, a dataset consisting of ~9 million URLs ... having a Creative Commons Attribution license* .<p>Then the footnote below:<p>&gt;* While we tried to identify images that are licensed under a Creative Commons Attribution license, we make no representations or warranties regarding the license status of each image and you should verify the license for each image yourself.<p>I think this might be the most blatant instance I&#x27;ve ever seen of, &quot;We have to write this even though it&#x27;s essentially impossible for you to actually follow our directions.&quot;

Interesting that the base data consists of URLs. I guess it makes sense given copyright issues. Anybody know what the ballpark expected half-life of such URLs?

Any guesses on how large the resulting dataset would be if you actually downloaded all the images? I imagine the urls will get removed in a hurry as everybody starts automating it.

First video, now images - wonder if speech and others are on the way?<p>It&#x27;s nice that they&#x27;re doing this, helps advance the art I think. But it also puts a lot of smaller operations in unis sort of under the Google system in that they&#x27;re best compared to Google&#x27;s ML work and others using these datasets. It&#x27;s a small way of stacking the deck to make Google and DeepMind more embedded in the community.<p>That said, its utility for others surely outweighs the strategic advantage gained here, so I for one welcome these libraries. A lot of work goes into them. Hopefully others will release theirs as well.

I&#x27;m glad I&#x27;m getting a return on all the effort clicking street signs and store fronts on reCaptcha.

I&#x27;ve put an efficient downloader here for the interested crowd: <a href="https:&#x2F;&#x2F;github.com&#x2F;beniz&#x2F;openimages_downloader" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;beniz&#x2F;openimages_downloader</a> It&#x27;s a fork of the one script I used to grab Imagenet.

Is there a link to the trained model somewhere?

Are there any other libraries that are similar?

Looking forward to someone trying tensorFlow CNN on this