DemoBias: An Empirical Study to Trace Demographic Biases in Vision Foundation Models
This work addresses fairness issues in biometric systems for diverse demographic groups, but it is incremental as it applies existing evaluation methods to new models and data.
The study tackled demographic biases in large vision-language models for face recognition by evaluating three models on a balanced dataset, finding that PaliGemma and LLaVA showed higher disparities for Hispanic/Latino, Caucasian, and South Asian groups, while BLIP-2 performed more consistently.
Large Vision Language Models (LVLMs) have demonstrated remarkable capabilities across various downstream tasks, including biometric face recognition (FR) with description. However, demographic biases remain a critical concern in FR, as these foundation models often fail to perform equitably across diverse demographic groups, considering ethnicity/race, gender, and age. Therefore, through our work DemoBias, we conduct an empirical evaluation to investigate the extent of demographic biases in LVLMs for biometric FR with textual token generation tasks. We fine-tuned and evaluated three widely used pre-trained LVLMs: LLaVA, BLIP-2, and PaliGemma on our own generated demographic-balanced dataset. We utilize several evaluation metrics, like group-specific BERTScores and the Fairness Discrepancy Rate, to quantify and trace the performance disparities. The experimental results deliver compelling insights into the fairness and reliability of LVLMs across diverse demographic groups. Our empirical study uncovered demographic biases in LVLMs, with PaliGemma and LLaVA exhibiting higher disparities for Hispanic/Latino, Caucasian, and South Asian groups, whereas BLIP-2 demonstrated comparably consistent. Repository: https://github.com/Sufianlab/DemoBias.