high-resolution-earth-geo-observatory-system-for-sentinel-2-imagery-using-deep-learning-hr-mars

Problem Statement:

A cost-effective statistical model is required which can predict super resolved landcover maps given sentinel-2 satellite imagery. These predicted maps are crucial to gain insights about Earth’s geographical changes and for monitoring the natural resources required to cope up with the alarmingly increasing population growth.

Proposed Solution:

An improved model to predict high resolution land cover for sentinel-2 remote sensing satellite image through super resolution. Easily accessible interface that allows users to generate statistical information about any selected geographical region in a web-based solution. #Contributions:

Generating updated highly resolved land cover without using costly conventional techniques. Obtained results perform even better than the satellite captured land covers because of 500meter resolution. Statistical analysis of International Geosphere Biosphere Program IGBP scheme. Results will improve forest change analysis, water level detection, climate change, crops monitoring and other areas using predicted super-resolved landcover information.

For detailed explaination see the presentation folder