The CNN model ( Figure 8b ) has an overall trend similar to CNN-GRU, but its Moran’s I and SSIM are significantly lower than those of the CNN-GRU model, and there are slight ßalt-and-pepper” noise artifacts in some localized areas, making the spatial smoothness slightly weaker.
← all excerpts
Improved leaf area index estimation in Jujube trees by fusing spatial-temporal features from UAV RGB time-series with a deep learning framework.
1
—
—