@Scp-3125  
Could be pretty useful to… make more SFW filtering content…
 
My, 2 years already, I now lean towards data visualization rather than to train the model lately.

I wonder how a visual transformer network would handle a problem like this. I imagine that it’s use of attention mechanisms would make it easier for it to spot porn

@Ganondox  
At the bottom you see Equal Error Rate 9.48% (point when false positive rate = false negative rate). Indeed at this point the filter would not be the most accurate thing ever. However without more information we can’t say how well it would work if you set it very sensitive to detecting explicit images (we need ROC curve for that).

What’s the false positive and false negative rate? Because I’m not going to trust a filter for explicit content that only has %89 accuracy.

Now that the subject I am studying can even be applied to manage pony porn.
 
This is hilarious!

Better than Tumblr’s dumbass bots.

This is amazing work! It would be very complimentary with the AI pony name generator.

Oh hey, that’s my penis!

@Background Pony #D6DA
 
Ah yes, my bad! I thought I was commenting on the [suggestive] version of the image ^^.
 
Yes, it is a non-explicit image which got high score for explicit. tag: "explicit" means the saliency map is computed for tag [explicit]. Admittedly it is rather suggestive image, which is probably why it got high score. This image was not in the training set.

@Background Pony #CE34
 
That’s not what I meant, although I may have misunderstood. >>1748971, here on Derpibooru, is not tagged “explicit.” In the image above, you say ‘tag: “explicit”’ underneath it. Do you mean that this is a non-explicit image which the program improperly thought was explicit? I just want to make sure a teaching image didn’t get miscategorized.

this is beautiful!

@dashbug  
Much like #9793 said, that model is not even aware of clop images and hence gives nonsense with such images.
 
I run this model here on the images you posted, and got 99% and 61% “explicit-percentage”.
 
(Sidenote: Ponies could make more interesting nsfw-detection dataset, because ponies come in different colors and shapes. Last time I checked, isitporn.com seems to focus on naked-skin color. We would rather want it to find explicit bits, much like this model did with the cherry-picked examples)

@dashbug  
That’s because it’s trained to detect IRL porn, not clop.

O.O?

>5% of the images (18.5k) were kept for evaluation (not used for training)  
I have such the best excuse if anyone finds my stash now

I can’t agree with you. That site doesn’t work correctly.  
Examples:  
http://isitporn.com/show/9a5c12fa91a7400d9d5ecaa8a5cdbe2d  
http://isitporn.com/show/d9ae26db4104466f89eb64d54f5312b7  
These explicit images are believed to be safe.
 
… But my cat images (even with only visible face) are almost nsfw

Hey my art’s there 👋

Gwern.net in the citations? Oh god, the streams are crossing again.

@Background Pony #D6DA  

 
Indeed it isn’t. I first aimed to tag this [safe] (hence the censor), but ultimately sticked with [suggestive] but kept the censor-block ^^’

@The Luna Fan  
This is almost as impressive as that AI that writes commercials for Olive Garden.

But >>1748971 isn’t tagged explicit?

Perfect, an AI that looks for clop, humanity is at it’s pick now!