Management of deep learning models for oil infrastructure extraction from remote sensing imagery

In remote sensing imagery, the extraction of oil infrastructure mainly relies on deep learning technology. However, previous studies have mostly focused on establishing and optimising models for oil infrastructure extraction, while neglecting the lifecycle management of training data, models, and experimental results. To facilitate research on oil infrastructure extraction from remote sensing imagery, this study establishes an Oil Infrastructure Data Management System (OIDMS) that integrates sample data organisation for oil infrastructure extraction with deep learning model. In this study, management is conducted on the model data for oil infrastructure extraction and the sample data used for model training. This involves establishing organisational models for oil infrastructure extraction samples and metadata for oil infrastructure extraction models. Furthermore, encoding mechanisms are designed to encode oil infrastructure extraction models, reducing the time of managing the platform for oil infrastructure extraction models and thereby improving the efficiency of model optimisation and management.