README.md

metadata

license: apache-2.0
task_categories:
  - text-to-image
language:
  - vi
tags:
  - text-rendering
  - ocr
  - synthetic-data
  - generative-models
  - fine-tuning
size_categories:
  - 100K<n<1M
pretty_name: Vietnamese Text Render 500K
configs:
  - config_name: complex
    data_files:
      - split: train
        path: data/complex/*
  - config_name: simple
    data_files:
      - split: train
        path: data/simple/*

Vietnamese Text Render 500K Dataset

A large-scale dataset containing 500K Vietnamese text rendering image-text pairs for training generative models to improve text rendering performance.

Dataset Structure

• image: Rendered text image in PNG format
• text: Corresponding text content
• filename: Original filename

Usage

This dataset is designed for fine-tuning generative models to improve text rendering capabilities on Vietnamese language.

from datasets import load_dataset
from PIL import Image
from io import BytesIO

dataset = load_dataset("pixxu/ViTextRender-500K") # Whole dataset include simple & complex images
print(dataset)

# Access a sample
sample = dataset['train'][0]
print(sample['text'])
img = Image.open(BytesIO(sample['image']))
img.show()

# If you run on Google Colab Notebook or Kaggle Notebook:
# img = Image.open(BytesIO(sample['image'])).convert("RGB")
# display(img)

If you want to load only simple or complex images:

from datasets import load_dataset
from PIL import Image
from io import BytesIO

dataset = load_dataset("pixxu/ViTextRender-500K", name="simple", split="train") # For simple images
# For complex images, use: dataset = load_dataset("pixxu/ViTextRender-500K", name="complex", split="train")
print(dataset)

# Access a sample
sample = dataset[0]
print(sample['text'])
img = Image.open(BytesIO(sample['image']))
img.show()

# If you run on Google Colab Notebook or Kaggle Notebook
# img = Image.open(BytesIO(sample['image'])).convert("RGB")
# display(img)

Applications

• Fine-tuning text-to-image models for better text rendering on Vietnamese language.

License

Apache License 2.0