Deep Learning for interpretation of subsurface radar data
|Coordinator||GUIDELINE GEO AB (PUBL)|
|Funding from Vinnova||SEK 498 000|
|Project duration||November 2018 - December 2019|
|Venture||Banbrytande idéer inom industriell utveckling|
|Call||Banbrytande idéer inom industriell utveckling - 2018|
Purpose and goal
Our goal in this project is to develop and apply Deep Learning techniques as a novel means of interpreting subsurface radar data, especially focused on buried infrastructure. Some examples include: power and signal cables, optical fiber, water and sewer pipes. We aim to develop a software library that will form a basis for further development and future commercialization of the technique.
Expected results and effects
Deep learning algorithms can automate and simplify the interpretation of subsurface radar data. Maintenance and installation of buried infrastructure can be performed cheaper and quicker. Increased use of radar technology can help create better digital maps and infrastructure databases. This will help owners and maintenance personell isolate and remediate problems, thus avoiding injury to people and property.
Planned approach and implementation
Work will be performed i four main areas: 1) create datasets suitable for training of the networks, 2) develop and optimize methods for supervised and unsupervised learning on subsurface radar data, 3) develop a reference implementation based on results from step 2, and 4) build software library based on results from step 3. The main deliverables are: 1) A reference implementation of the selected algorithms 2) A software library encapsulating the reference implementation 3) A brief overall project report