ResearchBenchmarksMedical Visual Question Answering on VQA 2019 (test)For AI assistantsMarkdownContent indexMedical Visual Question Answering on VQA 2019 (test)Follow79.87Overall AccuracyM3AE65.757269.421173.08576.7489Sep 15, 2022Apr 9, 2023Nov 2, 2023May 27, 2024Dec 19, 2024Jul 14, 2025Feb 6, 2026Evaluation resultsMethodMethodLinksOverall AccuracyM3AEvision encoder=CLIP-Vi...vision encoder=CLIP-ViT-B, language encoder=RoBERTa-base2022.0979.87MEVF-BAN2022.0977.86Robust-MMREvaluation Setting=Per...Evaluation Setting=Perturbed2026.0276.8CPRDEvaluation Setting=Per...Evaluation Setting=Perturbed2026.0269.6MEVF-SAN2022.0968.9MEVFEvaluation Setting=Per...Evaluation Setting=Perturbed2026.0268.1BANEvaluation Setting=Per...Evaluation Setting=Perturbed2026.0266.3