Benchmark tests whether AI can judge systems helping blind users

Original: A Visually Impaired Assistance Benchmark for VLM-as-a-Judge Evaluation

Source: arxiv.org ↗