4 ms·
Maybe start with just male genitalia. I hear some males have the tendency to send this unsolicited to females. It would be easy to build a data set around this.
by CTDOCodebases 4y ago
Maybe start with just male genitalia. I hear some males have the tendency to send this unsolicited to females. It would be easy to build a data set around this. Just put out an open call on social media. Be sure to specify 18+ and log all users submitting.
- senectus1 4y agohahaha. I can see a future of unsolicited non-existent dick pics being sent to people. for that matter, the catfishing aspect for thirsty people make this an interesting evolution. What a strange and fascinating world we're building for ourselves here.
- wongarsu 4y agoThat kind of catfishing is a good example why ethics review boards exist.
- myself248 4y ago"The image classifier said this image is 97% likely to be a dick pick, and has automatically muted it. Click here to show anyway. Click /here/ to keep it hidden but reply with a machine-generated bigger one."
- CTDOCodebases 4y agohttps://www.vice.com/en/article/y3dyax/chatroulette-moderation-penis-problems https://www.vice.com/en/article/y3dyax/chatroulette-moderati... Someone email this guy. We cracked it.
- knicholes 4y agoSomeone has to label the dick pics first, though.
- komali2 4y agoI volunteer as tribute
- knicholes 4y agoTo label the dicks or to have your dick labeled? :)
- dhosek 4y agoHot dog ⇔ Not hot dog
- knicholes 4y agoThat's a way to classify whether or not the pics are indeed dick pics, but the problem lies in how these recent diffusion models work. Typically you have two encoders, one for the text prompt and one for the image. Then you train a model to optimize where the prompt embeddings match the image embeddings. You've got to actually have some decent descriptions of the dick, like: thin, girthy, 7", 8", 4", micropenis, veiny, limp, hard, huge tip, bifurcated, foreskin, black, white, latino, asian, etc, if you want to generate with any accuracy your desired fake penis.