Skip to content

Evolving images of pokemon using pix2pix, and their names using a char-level seq2seq model.

Notifications You must be signed in to change notification settings

Hamcastle/Neural-Pokemon-Evolution

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Neural-Pokemon-Evolution

A project to "evolve" images of Pokemon using pix2pix, and their names using a char-level seq2seq model. The goal is to make predictions for the new Sword and Shield starters, then see how well I did when their evolved forms are released.

The predicted names are:

Sobble → Grobtil → Golodod

Scorbunny → Honun → Sullox

Grookey → Bolodom → Giglilic

I'm not happy with the quality of the predicted "evolved form" images as they are now: they're smeary and even with lots of training and in the best cases they end up looking like something out of The Fly. But just so you can see that this is in fact an exectuted-upon part of the project, the predictions for the new starters are below. Note that I still have to re-run the process on the second-evolution output forms, so those aren't available for the moment:

Sobble Grobtil Golodod
???
Scorbunny Honun Sullox
???
Grookey Bolodum Giglilic
???

And some cherry-picked examples from the test-set where it doesn't look like the network just put its hands inside the poor wee beasties and twisted its fingers around:

Meowth Persian Target Persian Prediction
Ratata Raticate Target Raticate Prediction

About

Evolving images of pokemon using pix2pix, and their names using a char-level seq2seq model.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published