Retina optical coherence tomography image detection method, device and terminal

The invention discloses a retina optical coherence tomography image detection method and device, a terminal and a medium, and the method comprises the steps: obtaining a plurality of image features of different scales through a convolutional neural network, and respectively extracting a local classification mark and a global classification mark through a Transform layer; according to the method, a global classification mark and a local classification mark are combined, a cross attention module is adopted in a Transform layer to exchange feature information among image features of different scales so as to make up the defect that a Transform model extracts local features, and then the global classification mark and the local classification mark are fused, that is, the local information and the global information in the image are further combined, so that an accurate recognition and classification result is obtained. 本发明公开了一种视网膜光学相干断层扫描图像检测方法、装置、终端及介质，通过卷积神经网络获取到不同尺度的多个图像特征，分别用来提取局部分类标记和通过Transformer层提取全局分类标记，并