DeepSight

• DeepSight is free web service that provides land use and land cover semantics of satellite images.
• It uses supervised learning to do so.
• DeepSight uses a convolutional neural network - SpectrumNet to automatically classify / label images obtained from the Sentinel-2 satellite. This builds an efficient and more effective method for land use planning.
• The underlying algorithm of these models is convolutional neural network (CNN).
• These models are provided to the users via a web service.
• The models can be accessed by the users via APIs. The user sends a satellite image and DL model is applied to this image and the land use / land cover data is returned to the user.
• SpectrumNet is a standardized, consistent and accurate classification system containing systematic and strict class boundary definitions, using remote sensing techniques to map land cover and land use from a series of multi-spectral satellite images.