MicroGlam✨: Microscopic Skin Image Dataset with Cosmetics

Toby Chong, Alina Chadwick, I-Chao Shen, Haoran Xie, Takeo Igarashi

MicroGlam is a dataset of skin patch images from 45 patches (5 skin patches each from 9 participants) of size 8mm $^*$ 8mm under three cosmetic products (i.e., foundation, blusher, and highlighter) under a variety of light conditions.

Download

Kaggle

Dataset Content

Images are encoded to HEIF (.heic) to minimize storage requirement. We recommend directly reading the image as is using library such as pyheif.

Folder/Filename Convension

After uncompressing the zip, you will find patch folders named S{}_P{}_{product_name}_{handiness}.

• S stands for subject number, i.e., all S1 folders come from the same person;
• P stands for patch number, i.e., all S1_P1 folders come from the exact same skin patch;
• product_name refers to the product applied onto the skin patch, and
• handiness refers to which hand the patch comes from.

In each of the patch folders, we have subfolders 1~8, the number refers to the number of LED light that is lit during the capture.
All images are named calibration_{}_{}_{}_{}_{}_{}_{}_{}_{}_{}_{}_{}_{}_{}_{}_{}.heic, Each of the 3 digit numbers refer to the intensity of the LED light (0~200). The lighting condition are randomized amoung subjects, but consistent for all patches for the same subject across all patches. The relative position of the LED light with regard to the camera is fixed.

License

This data is free to use for non-commercial academic research.

Contact

Should you have any question please feel free to open an issue and for commercial enquiry please contact me at tobyclh@gmail.com.