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

Figure 5

Comparison of the prediction accuracy between the fused feature-based encoding model and (a) VGG16-based or (b) FCN32-based encoding models in five visual areas. The ordinate represents the prediction accuracy of the fused feature-based encoding model, and the abscissas represent the prediction accuracy of the VGG16-based or FCN32-based encoding models. The orange dots represent the voxels that can be better predicted by the fused feature-based model than the VGG16-based or FCN32-based models. The blue dots represent the opposite. The green dashed lines and the black dots represent the same meanings as Figure 2. (c) Distribution of the difference between fused features and VGG16 or (d) FCN32-based model in prediction accuracies. The blue color denotes that the prediction accuracy is higher for the VGG16-based model or FCN32-based model. The orange color denotes that the prediction accuracy is higher for the fused feature-based model. The numbers on each side indicate the fraction of voxels with higher prediction accuracy under the model.

(a)

(b)

(c)

(d)