TextSculptor Data: Image-Text Pairs for Scene Text Editing

TextSculptor Data contains two Parquet subsets for scene text editing tasks. The subsets include columns for text captions or prompts paired with images stored as embedded bytes. The dataset is associated with a research project and was last updated on 2026-05-21.

Use Cases

• Training scene text editing models based on the paired image-text data.
• Benchmarking text rendering quality based on the 'text_rendering_qwen_image' subset.
• Generating synthetic scene text images based on the 'long_text_rendering' subset.
• Developing multimodal AI models that understand text within visual contexts.

Strengths

• The dataset is structured into two distinct subsets for specific tasks.
• Images are stored as a Hugging Face image feature with embedded bytes, which likely simplifies loading.

Limitations

• The number of rows, columns beyond those named, and total size are unknown.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Freshness

Last updated 2026-05-21 10:40:04.