Sunny Skin And Sunscreen Extract 1K: Skin Lesion and Packaging Images

A multimodal dataset contains 1,000 images of skin lesions sourced from the ISIC-2024 dataset and approximately 100 images of sunscreen packaging photographed in Australia. The dataset was created by author mrdbourke and last updated on February 19, 2026. It includes extractions of details from each image generated by the Gemini-3-Flash-Preview model.

Use Cases

• Fine-tuning vision-language models like MedGemma-1.5 based on the described image-to-text workflow.
• Training models to extract structured information from medical dermatology images.
• Developing models to read and interpret text and graphics on consumer product packaging.
• Creating educational or diagnostic aids for dermatological conditions based on the image collection.

Strengths

• Includes 1,000 skin lesion images from a known, permissively licensed source (ISIC-2024).
• Contains approximately 100 images of real-world sunscreen packaging.
• Features machine-generated extractions from images using a specific AI model (Gemini-3-Flash-Preview).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the extracted text data is unknown, which may limit suitability assessment.
• Data may reflect geographic bias inherent to the Australian source of the packaging images.

Provenance

Source

ISIC-2024 dataset and original photography by the author.

Collection Method

Images were sourced from ISIC-2024 and photographed; details were extracted via the Gemini-3-Flash-Preview AI model.

Freshness

Last updated 2026-02-19 02:38:41; freshness should be verified.

Geography

Sunscreen packaging images were taken in Australia.