Skip to main content
Fig. 4 | EURASIP Journal on Image and Video Processing

Fig. 4

From: A semi-supervised convolutional neural network based on subspace representation for image classification

Fig. 4

Conceptual figure of DFSNet. The network employs two distinct filters banks based on PCA and GDS. To reduce the high dimensionality of the feature vectors and increase rotation invariance, the proposed method is followed by a feature mapping that includes binarization and block-wise histogram. Similar to most shallow networks, the classification is performed by linear SVM

Back to article page