Comic 9K
Comic-9K is a dataset of images from comic books, combined with synopses written by humans. It is intended for research in computer vision, multimodal comprehension, and text generation.
Approximately 238,000 images in standard formats, accompanied by text synopses, Parquet files and tar.gz archives
Apache 2.0
Description
The dataset Comic-9K contains approximately 238,000 images taken from comic books, with human synopses describing the associated storylines. Data is provided in Parquet and compressed archive formats.
What is this dataset for?
- Train computer vision models specialized in comic books
- Develop multimodal models combining image and text (summaries, narrative generation)
- Explore the understanding and analysis of graphic and narrative content
Can it be enriched or improved?
Yes, the dataset can be enriched with additional annotations such as dialogues, case segmentation, or genre labels. It is also possible to extend the synopses or to add cultural metadata.
🔎 In summary
🧠 Recommended for
- Computer vision researchers
- Multimodal projects
- Graphic narrative content analysis
🔧 Compatible tools
- PyTorch
- TensorFlow
- Hugging Face Datasets
- Image analysis tools
💡 Tip
Use the synopsis structure to guide multimodal text generation models.
Frequently Asked Questions
What is the nature of the synopses in Comic-9K?
They are summaries written by humans that describe the plots associated with comic book images.
Does the dataset only include images or also text?
It includes extracted images as well as textual synopses, making it a multimodal dataset.
Can Comic-9K be used for text generation projects?
Yes, it is adapted to the generation and understanding of text in connection with visual content.
