DeepSentinel

DeepSentinel是由牛津大学环境与地理学院企业与环境史密斯学院开发的，用于Sentinel-1和Sentinel-2影像的传感器融合研究的数据集。该数据集包含100000条数据，涵盖全球范围内的土地覆盖信息，旨在通过机器学习技术解决气候变化带来的金融风险评估问题。数据集创建过程中，采用了自监督预训练和微调策略，以提高模型在特定应用中的性能。DeepSentinel的应用领域广泛，包括但不限于碳排放估算、甲烷排放监测和大型固定资产定位等。DeepSentinel is a dataset developed by the Smith School of Enterprise and the Environment, School of Geography and the Environment, University of Oxford, for sensor fusion research on Sentinel-1 and Sentinel-2 satellite imagery. This dataset contains 100,000 records covering global land cover information, and is designed to address financial risk assessment challenges arising from climate change via machine learning technologies. Self-supervised pre-training and fine-tuning strategies were employed during the dataset development phase to improve the model's performance for targeted applications. DeepSentinel supports a broad spectrum of application domains, including but not limited to carbon emission estimation, methane emission monitoring, and large fixed asset localization.