Does your satellite data suffer from sub-optimal resolution? Find out how InCubed activity in1 can be used to enhance your satellite data at Sobolt’s free two-week virtual trial, starting on Monday 15 March 2021.
AI solutions provider Sobolt joined the Investing in Industrial Innovation (InCubed) programme in February 2020 and has worked to develop state-of-the-art Artificial Data Enhancement (ADE) software called in1.
in1 uses a sophisticated image processing model for unpaired super-resolution to reduce the cost of enhancing satellite data in real-time. The so-called “unpaired” problem consists in super-resolving a low resolution satellite image with AI deep learning techniques, without having a high resolution target image to compare it to. To do so, Sobolt has developed a dedicated state-of-the-art artificial intelligent model that provides potential customers with upgraded data and unlocks exciting new use cases for existing datasets and satellite programmes such as more accurate data analyses. Find out more here.
The free Copernicus Sentinel-2 trial of in1 will start virtually on Monday 15 March 2021 and run for two weeks. Anyone working with Sentinel-2 images can participate in the free trial, using Sobolt’s image library provided on the website or super resolving own images with an API in the cloud. In three easy steps participants will be able to upload an image, enter a short line of code into the API, and finally click on a button to get a higher resolution Sentinel-2 image emerging in a matter of seconds.
This free trial will allow participants to get first-hand experience of in1, be able to evaluate the models performance, and provide valuable feedback. Please register by emailing email@example.com – hurry, only 19 spots available!
Since in1’s intermediate milestone meeting as part of the InCubed programme in December 2020, in1 has been enhancing the sharpness of Copernicus Sentinel-2 images whilst preserving and exceeding Sobolt’s benchmarking for bilinear upsampling. The meeting focused on performance metrics of in1’s original colour definition and how well object edges are defined such as houses and trees.
In1 is a toolbox of AI deep learning functionality focused on improving and enhancing Earth Observation imagery and data sets. This is an innovative software development which pushes the architectural boundaries of deep learning techniques and GPU operational capability. Using a continual improvement approach, Sobolt have developed automatic testing procedures and conducted external peer reviews on every new piece of code added to consistently improve the toolbox.
When: Monday 15 March 2021
To know more and register: Email firstname.lastname@example.org