Shulou(Shulou.com)11/24 Report--

CTOnews.com, September 2 (Xinhua)-- in order to alleviate the systematic bias against women and people of color in many computer vision models, Meta recently launched a new AI tool called FACET, which is used to identify racial and gender biases in computer vision systems.

The FACET tool is currently trained with 30,000 images and 50,000 images, especially to enhance the perception of gender and skin color, and can be used to evaluate various features of computer vision models.

The FACET tool can be trained to answer complex questions, such as identifying male subjects and further identifying skateboarders, as well as light and dark skin.

Meta uses FACET to evaluate the DINOv2 model and SEERv2 model developed by the company, as well as OpenAI's OpenCLIP model. On the whole, OpenCLIP performs better than other models in terms of gender, while DINOv has better judgment in terms of age and skin color.

The open source of FACET will help researchers perform similar benchmarking tests to understand the biases in their models and to monitor the impact of mitigation measures taken to address fairness issues. CTOnews.com attaches the address of the Meta press release here, which can be read in depth by interested users.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.