Applying Knowledge Transfer for Water Body Segmentation in Peru.

摘要: In this work, we present the application of convolutional neural networks for segmenting water bodies in satellite images. We first use a variant U-Net model to segment rivers and lakes from very high-resolution images Peru. To circumvent issue scarce labelled data, investigate applicability knowledge transfer-based that learns mapping combines it with so better segmentation can be achieved. train single process, end-to-end. Our preliminary results show adding information available does not help out-of-the-box, fact worsen results. This leads us infer data could different distribution, its addition increased variance our