以下内容由机器翻译生成。如果您觉得可读性不好, 请阅读原文或 点击这里.
Pinterest has rolled out new changes that let users filter hair and beauty search results by skin tone ranges, as a way to make it easier to find what you may be looking for. The changes come after the social media platform conducted a survey and found that 70 percent of people used the service to find and save everyday looks and styles.
A Medium post from Pinterest’s engineering account says the process of programming the search filters to detect skin tone in images was difficult, given how lighting, shadows, and other factors were all variables to consider.
To solve the issue, Pinterest used machine learning from a third-party beauty app ModiFace that already had a suitable facial data library for artificial intelligence use. Pinterest then trained the algorithms not to read white skin in dark shadow as dark skin tone, and corrected other mistakes through repeated testing.
Pinterest also noted that when users searched for skin tones, that information wouldn’t be stored and it wouldn’t affect ads served to users. The feature is rolling out today in beta, the company says.
“Our future work will generally focus on improving the accuracy of results and bringing the experience to more platforms. We’ll try out new methods of query rewriting and blending in actionable content and hope to improve our search ranking models in order to better take into account the selected skin tone,” writes Laksh Bhasin, a Pinterest engineer working on its search quality team. “Lastly, skin tones are just the start of building a more inclusive search. We hope to help Pinners find more personalized results by offering more ways to narrow your search.”