Haut.AI gets patent for Skin Atlas system
Haut.AI, a skin analysis technology company, has been granted a patent in the United States for its Skin Atlas system.
Skin Atlas, which removes personal identifiers and non-skin-related data from facial images, aims to help ensure privacy and security while also helping maintain analytical accuracy in e-commerce and clinical skin analysis applications.
The tool works by having users take a high-quality image that is sent to the cloud for processing, where the company’s anonymization technique strips the facial images of all personal identifiers and leaves only essential skin pixels behind for analysis, the company said.
[Related: Skincare Stories]
“‘Skin Atlas’ enables users to confidently share their images for analysis, knowing their personal information is protected. The anonymization process ensures that all identifying features, such as eyes, mouth, hair and background details, are removed, focusing solely on the skin. This safeguards privacy and also enhances the accuracy of our skin assessments by eliminating potential biases from non-skin-related information. We are excited about the potential of this technology to provide more secure and accurate skin diagnostics, ultimately benefiting both users and professionals in the skincare industry” explains Anastasia Georgievskaya, CEO of Haut.AI.
As a way to help data privacy standards, Skin Atlas looks at facial key points to standardize the image, followed by the separation of skin pixels from non-skin pixels, which is crucial for accurate analysis. The system then generates photo-realistic skin patterns to fill the non-skin areas, creating a fully anonymized image. This process eliminates irrelevant data, such as hair and background details, ensuring that only pertinent skin information is analyzed. Additionally, the system includes advanced algorithms for segmenting and reconstructing the skin regions, the company said.
In the e-commerce sector, it enables personalized product recommendations without compromising user data.
