Vision-Language Models under Cultural and Inclusive Considerations
This work addresses the need for more inclusive AI assistants for visually impaired people across different cultures, though it is incremental as it builds on existing datasets and evaluation methods.
The researchers tackled the problem that current vision-language models may not adequately serve culturally diverse visually impaired users by creating a culture-centric evaluation benchmark from VizWiz data. They found that while state-of-the-art models show promise, they face challenges like hallucination and misalignment with human judgment.
Large vision-language models (VLMs) can assist visually impaired people by describing images from their daily lives. Current evaluation datasets may not reflect diverse cultural user backgrounds or the situational context of this use case. To address this problem, we create a survey to determine caption preferences and propose a culture-centric evaluation benchmark by filtering VizWiz, an existing dataset with images taken by people who are blind. We then evaluate several VLMs, investigating their reliability as visual assistants in a culturally diverse setting. While our results for state-of-the-art models are promising, we identify challenges such as hallucination and misalignment of automatic evaluation metrics with human judgment. We make our survey, data, code, and model outputs publicly available.