Meta has launched FACET (Fairness in Computer Vision Evaluation Tool), a tool to identify and rectify racial and gender biases inherent in computer vision systems.
The tool is based on Meta’s extensive dataset of over 30,000 images featuring 50,000 individuals, meticulously tagged by experts across diverse categories. This allows researchers to evaluate computer vision models’ performance concerning various characteristics, including perceived gender and skin tone.
FACET opens the door to crucial questions about the fairness of AI systems. For example, does an AI system excel at identifying male skateboarders? Does it perform better in recognizing individuals with specific skin tones? Moreover, does the system’s efficiency vary when faced with individuals possessing curly or straight hair?
In addition to launching FACET, Meta is also re-licensing its DINOv2 computer vision model under the Apache 2 open-source license, enabling commercial use. This move is a further commitment by Meta to fairness in AI, as it will allow other developers to build on DINOv2 and create more equitable AI systems.
Chloe Bakalar, Meta’s chief ethicist, emphasized the company’s dedication to advancing AI systems responsibly, especially concerning historically marginalized communities.
The sources for this piece include an article in Axios.