text stringclasses 1
value |
|---|
A digital_media_(illustration) depicting a split scene, one side shows a hooded figure walking towards a group of people in a cool-toned, graffiti-covered alley, the other side shows a lone figure hunched over in a warm-toned, dimly lit street, both scenes feature streetlights, implying different paths or choices, data... |
Processed Image Captions with Gemini Tags
This dataset contains a curated collection of processed images paired with captions generated using OpenAI's Gemini 1.5 Flash model. The captions combine structured tags and natural language, formatted as regular expressions to support text-to-image training workflows.
Key Features
- Caption Style: Captions include media-type tags (e.g.,
graphite_(artwork),digital_media_(photo),digital_media_(illustration)) and thematic descriptions inferred from the dataset's structure. - Generation Process: Captions were generated programmatically to combine concise descriptions with high-quality metadata, ensuring compatibility with machine learning tasks.
- Curated Content: The collection was curated and processed by the author to maintain quality and consistency.
Tags
- image-captioning
- text-to-image
- generative-ai
- gemini-model
- dataset-curation
- machine-learning
- Downloads last month
- 102