Sea Ice Segmentation From SAR Data by Convolutional Transformer Networks (2306.07649v1)
Abstract: Sea ice is a crucial component of the Earth's climate system and is highly sensitive to changes in temperature and atmospheric conditions. Accurate and timely measurement of sea ice parameters is important for understanding and predicting the impacts of climate change. Nevertheless, the amount of satellite data acquired over ice areas is huge, making the subjective measurements ineffective. Therefore, automated algorithms must be used in order to fully exploit the continuous data feeds coming from satellites. In this paper, we present a novel approach for sea ice segmentation based on SAR satellite imagery using hybrid convolutional transformer (ConvTr) networks. We show that our approach outperforms classical convolutional networks, while being considerably more efficient than pure transformer models. ConvTr obtained a mean intersection over union (mIoU) of 63.68% on the AI4Arctic data set, assuming an inference time of 120ms for a 400 x 400 squared km product.
- “Arctic Corridors and Northern Voices: Governing Marine Transportation in the Canadian Arctic (Pond Inlet, Nunavut Community Report),” 2018.
- “Guided Unsupervised Learning by Subaperture Decomposition for Ocean SAR Image Retrieval,” arXiv preprint arXiv:2209.15034, 2022.
- “Guided Deep Learning by Subaperture Decomposition: Ocean Patterns from SAR Imagery,” in Proceedings of IGARSS. IEEE, 2022, pp. 6825–6828.
- “Sea Ice and Open Water Classification of SAR Images Using a Deep Learning Model,” in Proceedings of IGARSS. IEEE, 2020, pp. 3051–3054.
- “Transferred Deep Learning for Sea Ice Change Detection From Synthetic-Aperture Radar Images,” IEEE Geoscience and Remote Sensing Letters, vol. 16, no. 10, pp. 1655–1659, 2019.
- “Semi-Supervised Sea Ice Classification of SAR Imagery Based on Graph Convolutional Network,” in Proceedings of IGARSS. IEEE, 2022, pp. 1031–1034.
- “Deep Learning for SAR Image Despeckling,” Remote Sensing, vol. 11, no. 13, pp. 1532, 2019.
- “An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale,” in Proceedings of ICLR, 2020.
- “Swin Transformer: Hierarchical Vision Transformer Using Shifted Windows,” in Proceedings of ICCV, 2021, pp. 10012–10022.
- “MeltPondNet: A Swin Transformer U-Net for Detection of Melt Ponds on Arctic Sea Ice,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 15, pp. 8776–8784, 2022.
- “Multi-Category Segmentation of Sentinel-2 Images Based on the Swin UNet Method,” Remote Sensing, vol. 14, no. 14, pp. 3382, 2022.
- “Exploring Vision Transformers for Polarimetric SAR Image Classification,” IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–15, 2021.
- “SatViT: Pretraining Transformers for Earth Observation,” IEEE Geoscience and Remote Sensing Letters, vol. 19, pp. 1–5, 2022.
- “Self-Supervised Pretraining of Transformers for Satellite Image Time Series Classification,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 14, pp. 474–487, 2020.
- “AI4Arctic Sea Ice Challenge Dataset,” Technical University of Denmark.
- “CVT: Introducing Convolutions to Vision Transformers,” in Proceedings of ICCV, 2021, pp. 22–31.
- “Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift,” in Proceedings of ICML. PMLR, 2015, pp. 448–456.
- “Rectified Linear Units Improve Restricted Boltzmann Machines,” in Proceedings of ICML, 2010.
- “Focal Loss for Dense Object Detection,” in Proceedings of ICCV, 2017, pp. 2980–2988.
- “U-net: Convolutional Networks for Biomedical Image Segmentation,” in Proceedings of MICCAI. Springer, 2015, pp. 234–241.