tailieunhanh - Application of Deep Learning Algorithm to Build an Automated Cloud Segmentation Model Based on Open Data Cube Framework

In this study, the method based on the proposed deep-learning method called ODC-Cloud, which was built on convolutional blocks and integrating with the Open Data Cube (ODC) platform. The results showed that our proposed model achieved an overall 90% accuracy in detecting cloud in Landsat 8 OLI imagery and successfully integrated with the ODC to perform multi-scale and multi-temporal analysis. | VNU Journal of Science Earth and Environmental Sciences Vol. 36 No. 4 2020 8-16 Original Article Application of Deep Learning Algorithm to Build an Automated Cloud Segmentation Model Based on Open Data Cube Framework Pham Vu Dong1 Bui Quang Thanh1 Nguyen Quoc Huy1 Vo Hong Anh2 Pham Van Manh1 1 VNU University of Science 334 Nguyen Trai Hanoi Vietnam 2 Central Remote Sensing Station National Remote Sensing Department 79 Van Tien Dung Tu Liem Hanoi Vietnam Received 11 September 2019 Revised 23 April 2020 Accepted 28 August 2020 Abstract Cloud detection is a significant task in optical remote sensing to reconstruct the contaminated cloud area from multi-temporal satellite images. Besides the rapid development of machine learning techniques especially deep learning algorithms can detect clouds over a large area in optical remote sensing data. In this study the method based on the proposed deep-learning method called ODC-Cloud which was built on convolutional blocks and integrating with the Open Data Cube ODC platform. The results showed that our proposed model achieved an overall 90 accuracy in detecting cloud in Landsat 8 OLI imagery and successfully integrated with the ODC to perform multi-scale and multi-temporal analysis. This is a pioneer study in techniques of storing and analyzing big optical remote sensing data. Keywords Optical Remote Sensing Landsat 8 OLI automatic cloud detection deep-learning Open Data Cube. ________ Corresponding author. E-mail address manh10101984@ https 2588-1094 8 . Dong et al. VNU Journal of Science Earth and Environmental Sciences Vol. 36 No. 4 2020 8-16 9 Ứng dụng thuật toán học máy sâu xây dựng mô hình tự động phát hiện vùng mây trên nền tảng dữ liệu khối Phạm Vũ Đông1 Bùi Quang Thành1 Nguyễn Quốc Huy1 Võ Hồng Anh2 Phạm Văn Mạnh1 1 Trường Đại học Khoa học Tự nhiên ĐHQGHN 334 Nguyễn Trãi Hà Nội Việt Nam 2 Đài Viễn thám Trung ương Cục Viễn thám Quốc gia 79 Văn Tiến Dũng Bắc Từ Liêm Hà Nội .