Generating Images from Chinese Text prompts using PaddleHub ERNIE-ViLG AI model
H
I build systems that blend AI and automation to solve real-world problems
While most of the popular AI image generation AI tools focus on English prompts, not many are in other languages.
ERNIE-ViLG is AI text to image generation model that is trained specifically for Chinese text to image generation.
ERNIE-ViLG is created by an organization called PaddlePaddle.
Table of Contents
- Using ERNIE-ViLG to generate images from Chinese text online
- Using ERNIE-ViLG in Google Colab Notebook
- PaddleHub ERNIE-ViLG Resources
Using ERNIE-ViLG to generate images from Chinese text online
The easiest way to start using ERNIE-ViLG to generate images from Chinese text is by visiting their demo on HuggingFace.
Using ERNIE-ViLG in Google Colab Notebook
You can also use ERNIE-ViLG in a Google Colab Notebook to generate images from Chinese text. Here are the steps:
- From the top menu bar choose Runtime -> Run All
- It will launch a Gradio app inside the Google Colab that you can use to generate images from Chinese text
- Gradio App can be opened in a separate URL and shared with others for 72 hours.
- You could also run ERNIE-ViLG locally on your machine by downloading this Colab Notebook and running it as Jupyter Notebook
PaddleHub ERNIE-ViLG Resources
The source code for ERNIE-ViLG is available here and the Paper can be read here