Young women often complain about the daily fluctuation of their facial skin conditions. However, no objective study has been carried out on such changes. This study is aimed at quantitatively elucidating daily skin fluctuation and evaluating the efficacy of cosmetic skin care treatment. We developed the first portable and self-guided facial skin imaging device (eMR Pro) to reproducibly capture facial images at home. Two 8 week clinical studies were then conducted to analyze daily skin fluctuation of facial pore areas, roughness and redness in young Japanese women (n = 47 in study 1 and n = 57 in study 2) by collecting facial images three times a day, during the morning after wake-up, during the morning after face wash, and during the evening after face wash. After a 4 week baseline measurement period (week -4 to week -1), all subjects applied Galactomyces ferment filtrate (GFF, Pitera^®) skin care formula twice a day for 4 weeks (week 1 to week 4). These three skin conditions did exhibit different fluctuation patterns. The pore area and roughness showed the “morning after wake-up”-largest fluctuation pattern, whereas redness showed the “evening after face wash”-largest fluctuation pattern. GFF treatment significantly reduced the net values and delta fluctuation of pore area, roughness, and redness, which were consistently observed in two studies. In conclusion, the daily fluctuation of facial skin conditions is potentially a new target field for investigating healthy skin maintenance.

Facial pore segmentation results can provide reliable evidence to simulate post-product pore conditions and provide product recommendations. However, accurately segmenting pores is challenging due to their small size, weak boundaries and dense distribution. It is also difficult to acquire precise annotation. Therefore, we formulate pore segmentation as a two-stage, weakly supervised task using both traditional and deep learning methods without human annotation. We propose a novel method called the pore segmentation network (PS-Net). Specifically, it contains pore feature extraction with coarse labels generated by a traditional method, as well as fine segmentation with progressively updated pseudo labels. Since pores provide high-frequency information about faces, we propose a high-frequency attention module that emphasizes low-level features. Moreover, we design a Bayesian module to identify pore shapes in high-level features. We establish a large-scale facial pore dataset with coarse labels that were generated via the difference of Gaussian (DoG) Pore method. PS-Net achieves the best performance on this dataset, proving its superiority compared with existing state-of-the-art segmentation methods.