BFENet: A two-stream interaction CNN method for multi-label ophthalmic diseases classification with bilateral fundus images

•We propose a two-stream interactive CNN architecture for multi-label ophthalmic diseases classification with bilateral fundus images.•We design a feature enhancement module by using the attention mechanism to learn the interdependence between local and global information, which enhances the extracted feature under the two-stream interactive architecture.•A multiscale module is designed to enriches the feature maps by superimposing feature information of different resolutions images extracted through dilated convolution.