A Weakly-Supervised, Multitask Deep Learning Framework for Shadow Mitigation in Remote Sensing Imagery
- Resource Type
- Conference
- Authors
- Couwenhoven, Scott D.; Ientilucci, Emmett J.; Park, Byung H.; Hughes, David
- Source
- IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium Geoscience and Remote Sensing Symposium, IGARSS 2022 - 2022 IEEE International. :619-622 Jul, 2022
- Subject
- Aerospace
Components, Circuits, Devices and Systems
Computing and Processing
Engineered Materials, Dielectrics and Plasmas
Fields, Waves and Electromagnetics
Geoscience
Photonics and Electrooptics
Power, Energy and Industry Applications
Signal Processing and Analysis
Training
Earth
Deep learning
Image segmentation
Histograms
Satellites
Inference algorithms
cloud shadow mitigation
deep learning
weak supervision
multitask learning
satellite imagery
- Language
- ISSN
- 2153-7003
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.