Press Release
Free Access

Table B.2.

Overview of the SCAN-magnetogram model.

Layer Convs Filters Sampling Input tensor shape Output tensor shape
Input 16 (1, 128, 128) (16, 128, 128)
ConvBlock 1 16 Down (16, 128, 128) (16, 128, 128)1; (32, 64, 64)
ConvBlock 2 32 Down (32, 64, 64) (32, 64, 64)2; (64, 32, 32)
ConvBlock 3 64 Down (64, 32, 32) (64, 32, 32)3; (128, 16, 16)
ConvBlock 3 128 (128, 16, 16) (128, 16, 16)
ConvBlock 3 64 Up (128, 16, 16); (64, 32, 32)3 (64, 32, 32)
ConvBlock 2 32 Up (64, 32, 32); (32, 64, 64)2 (32, 64, 64)
ConvBlock 1 16 Up (32, 64, 64); (16, 128, 128)1 (16, 128, 128)
Output 1 (16, 128, 128) (1, 128, 128)

Notes. The superscript number for the input- and output-tensors indicate the skip-connections. Tensor shapes are given in channels-first format. Convs refers to the number of convolutional layers in the ConvBlock.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.