Hyperspectral Skin Vision Challenge: Can Your Camera See Beyond Your Skin?

The ICASSP-SP Grand Challenge on Hyperspectral Skin Vision aims to democratize skin analysis by leveraging low-cost consumer-grade cameras to reconstruct vital spectral reflectance data. Addressing the accessibility limitation of costly hyperspectral equipment, this challenge tasks participants with decoding skin spectral information crucial for assessing melanin and hemoglobin concentrations. The provided Hyper-Skin dataset is carefully curated following ethical guidelines, consisting of a total of 306 hyperspectral data from 51 human subjects. Through comprehensive evaluation with Spectral Angle Mapper (SAM), fairness and accuracy in spectral reconstruction methods are ensured, encouraging advancements with real-world applications. This challenge attracted 51 teams and yielding 9 complete submissions. The top 5 teams achieves significant performance with average SAM score of 0.09486. This achievement underscores the transformative potential of this initiative in reshaping skin analysis accessibility and driving interdisciplinary progress with profound societal implications.