Skip to main content
Fig. 2 | BioMedical Engineering OnLine

Fig. 2

From: Automatic diagnosis of imbalanced ophthalmic images using a cost-sensitive deep convolutional neural network

Fig. 2

The architecture of the CS-ResCNN method. a The overall layers and connections of the CS-ResCNN model consisting of convolution layers, a max-pooling operation and 16 residual blocks, indicated by the red, green and blue rectangles respectively, followed by softmax and cost-sensitive adjustment layers. b One unfolded residual block is presented. c The BN and scale operations are presented. CS-ResCNN, cost-sensitive residual convolutional neural network; BN, batch normalization; Conv, convolution operation; ReLU, rectified linear unit

Back to article page