A Weakly-Supervised, Multitask Deep Learning Framework for Shadow Mitigation in Remote Sensing Imagery

Conference · Fri Jul 01 04:00:00 EDT 2022

DOI:https://doi.org/10.1109/IGARSS46834.2022.9883550· OSTI ID:2376405

Couwenhoven, Scott ^[1]; Ientilucci, Emmett ^[1]; Park, Byung ^[2]; Hughes, David ^[2]

---

1. Rochester Institute of Technology (RIT), New York
2. ORNL

We propose a weakly-supervised, multitask framework for training a convolutional neural network to solve the problem of cloud shadow mitigation given only cloud and shadow masks as labels. The network minimizes the Wasserstein distance between shadows and their proximal sunlit neighborhoods, generating a supervisory signal directly from within the input image. We extract further utility from the shadow mask through multitask learning by introducing an auxiliary task of shadow segmentation. Our approach is advantageous since it performs mitigation in an end-to-end framework which requires only a shadowed image for inference. We apply this process to the Landsat 8 OLI SPARCS validation data set and demonstrate plausible results.

• Research Organization: Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States)
• Sponsoring Organization: USDOE
• DOE Contract Number: AC05-00OR22725
• OSTI ID: 2376405
• Country of Publication: United States
• Language: English