*2.2. Flow Chart*

The flow chart of the proposed framework is presented in Figure 1.

The pairs of generated seismic and the corresponding mask compose the training dataset and feature label, as the input for the proposed U-Net neural network. The training procedure is shown in Figure 2.

**Figure 2.** Training procedure.

After the training process, the testing data (not included in training) will be input to the trained U-Net neural network, and the feature label set is predicted for verification purposes, which is shown in Figure 3.

**Figure 3.** Verification procedure.
