CoolFace
Datasetpublic

Nininkkka/Synth-Text-Eng-512x128

Synthetic Text Images (English) A synthetic dataset of rendered text images with rich per-sample annotations: the text itself, its rendering attributes, background description, applied post-processing, and a natural-language caption. Each image is generated by compositing English text over a procedurally generated background with random font, color, position, rotation, blur, brightness and noise. All samples are accompanied by a structured metadata.csv and a ready-to-use… See the full description on the dataset page: https://huggingface.co/datasets/Nininkkka/Synth-Text-Eng-512x128.

sourceHugging Facemitupdated 4h agoView on Hugging Face
0likes
Dataset Card

Synthetic Text Images (English)

A synthetic dataset of rendered text images with rich per-sample annotations: the text itself, its rendering attributes, background description, applied post-processing, and a natural-language caption.

Each image is generated by compositing English text over a procedurally generated background with random font, color, position, rotation, blur, brightness and noise. All samples are accompanied by a structured metadata.csv and a ready-to-use caption.

Dataset structure

data/
├── train/
│   ├── images/
│   │   ├── shard_0000/
│   │   │   ├── 000000.png
│   │   │   └── ...
│   │   └── shard_0001/
│   │       └── ...
│   └── metadata.csv
├── validation/
│   └── ...
└── test/
    └── ...

All images are 512 × 128 px, RGB, PNG.

Splits

SplitSize
train32000
validation4000
test4000

Metadata fields

ColumnTypeDescription
file_namestringRelative path to the image
textstringGround-truth text rendered on the image
text_lengthintNumber of characters in text
num_wordsintNumber of whitespace-separated words
fontstringFont family name (informational only)
font_sizeintFont size in pixels
text_colorstringCoarse color name of the text
text_color_hexstringHex RGB of the text color (#rrggbb)
bg_typestringOne of solid, gradient, noisy, textured
bg_colorstringDescription of the background color (may include hex endpoints for gradients)
rotation_degfloatApplied rotation in degrees (0 if none)
blur_sigmafloatGaussian blur σ (0 if none)
brightnessfloatBrightness multiplier (1.0 if none)
noise_ampintAmplitude of additive pixel noise (0 if none)
strokeint1 if a stroke/outline was applied to the text, else 0
stroke_widthintStroke width in pixels (0 if none)
text_x, text_yintPIL draw origin of the text (not the top-left of the visible bbox)
captionstringNatural-language description of the image

Usage

python
from datasets import load_dataset

ds = load_dataset("Nininkkka/Synth-Text-Eng-512x128")

# OCR-style access
sample = ds["train"][0]
print(sample["text"])          # ground-truth text
sample["image"].show()

# Captioning-style access
print(sample["caption"])

Intended uses

  • —Training and evaluating OCR / scene-text recognition models.
  • —Image captioning with fine-grained attributes (color, font size, background type).
  • —Research on text rendering and typography.
  • —Pretraining / data augmentation for text-in-image tasks.

Licensing

The dataset (images and metadata.csv) is released under the MIT License.

Images were rendered using fonts licensed under the SIL Open Font License 1.1. The font files themselves are not redistributed with this dataset.

Nininkkka/Synth-Text-Eng-512x128 · CoolFace