BPI-MVQA: a bi-branch model for medical visual question answering

BMC Medical Imaging

Table 14 Experimental comparison of BPI-MVQA with other state-of-the-art methods on the ImageCLEF2019 VQA-Med dataset

Model	Accuracy	BLEU
VGG16(GAP)+BERT+MFB [27]	0.624	0.644
ResNet152+BERT+Skip-thought vector [50]	0.616	0.634
ResNet152+LSTM+co-attention+MFH [51]	0.566	0.593
SFN [28]	0.558	0.582
GCVMVQA [52]	0.640	0.659
BPI-MVQA(Our model)	0.654	0.687

ISSN: 1471-2342