Seismic Facies Segmentation Using Ensemble of Convolutional Neural Networks

Figure 11

(a–e) Predictions by SegNet+augmentation, SegNet+no augmentation, DeepLabv3Plus+augmentation, DeepLabv3Plus+no augmentation, and ensemble. (f–j) Green and magenta colours show the difference between ground truth and prediction of the corresponding model from (a) to (e).

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)