Skip to main content
. 2024 May 1;24(9):2905. doi: 10.3390/s24092905

Figure 4.

Figure 4

The structure of SPPFCSPC_EMA. SPPFCSPC performs a series of convolutions on the feature map, followed by max-pooling and fusion over four receptive fields (one 3 × 3 and three 7 × 7). After further convolution, it is fused with the original feature map and finally combined with EMA to form the SPPFCSPC_EMA module (Conv: convolution; MaxPool2d: max pooling) [68].