Are there any ML engines that go from images to captions(inverse prompts?)
Would be interesting to prompt an image and then caption it
Conversation
Replying to
Oh this would be so good, I'd just keep feeding them back and forth until reaching a result equilibrium as with google translate swapping between languages.
1
3
Replying to
Basically having AI play paper telephone with itself:
1
Replying to
BLIPP (Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation) does this github.com/salesforce/BLIP but for reverse engineering prompts, CLIP Interrogator is better as others linked to.
1
2
Replying to
this colab notebook works reasonably well for this
Quote Tweet
fun fun!
You can reverse engineer the prompt from an image using this cool notebook by @pharmapsychotic
colab.research.google.com/github/pharmap
Show this thread
1
Show more replies










