https://github.com/pankajr141/gimmick

Its a image generation library which learns to generate patterns based on training data
https://github.com/pankajr141/gimmick

autoencoders computer-vision gan

Last synced: 4 months ago
JSON representation

Its a image generation library which learns to generate patterns based on training data

• Host: GitHub
• URL: https://github.com/pankajr141/gimmick
• Owner: pankajr141
• License: apache-2.0
• Created: 2021-06-06T07:27:19.000Z (about 5 years ago)
• Default Branch: main
• Last Pushed: 2021-10-19T11:27:02.000Z (over 4 years ago)
• Last Synced: 2025-09-25T12:16:08.326Z (10 months ago)
• Topics: autoencoders, computer-vision, gan
• Language: HTML
• Homepage:
• Size: 533 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# gimmick

## Introduction

Whats Gimmick ? Its a library which allows us to generate images out of nothing, basically you tell gimmick what kind of images you would like to generate and what model to use and it does the rest.

Its easy to use as primary focus is on user friendlyness without exposing to much details, but you can be specific if you wanna be.

What do we do internally ? we trained neuralnets, we add different type of networks making sure that interface always remain same, just plugin different algo and see how it perform.

## Installtion

pip install gimmick

just like that, its python everything should be easy


## Dependency

Gimmick autmatically install all the dependency for you, but occationaly you run into one of those bugs, so in a nutshell we use tensorflow, sklearn.

i know torch is much faster, we will get their

## Examples

we have already provided samples with the github repo, you can browse it, read it and run it.

Lets demonstrate

### usecase 1 - I just wanna run as simple as i can.


import gimmick

model = gimmick.learn(images, algo='autoencoder_dense')  # Just make sure that images is a numpy array which contains N number 3D or 2D images

images_gen = model.generate(16)  # I need 16 images



### usecase 2 - I wanna control how my model is trained


import gimmick

model = gimmick.learn(images, algo='autoencoder_dense', epochs=500, batch_size=8,

                      optimizer='adam', learning_rate=0.01, loss_function='mae')

images_gen = model.generate(16, batch_size=8)



### usecase 3 - I wanna know everything

Go through the detailed documentation.