A Comparative Analysis of Visual Encoding Models Based on Classification and Segmentation Task-Driven CNNs

<div>Relationship between the percentage of selected features and prediction accuracy of two encoding models in five visual areas: (a) VGG-based model and (b) FCN32-based model. The abscissa represents the percentage of selected features, and the ordinate represents the prediction accuracy of the models. The lines represent the results for five different visual areas.</div>

Computational and Mathematical Methods in Medicine

fig8

Figure 8

Figure 8: A Comparative Analysis of Visual Encoding Models Based on Classification and Segmentation Task-Driven CNNs 