Pin out objects manually and collaboratively.
Export training datasets for training deep learning algorithms.
Detect objects running a Target Detection algorithm over an area of interest.
Validate annotations made by other users.
Build datasets for object detection on satellite imagery.
OBJECTIVES
Traditional empirical and analytical Earth Observation (EO) algorithms are getting to a fundamental change, where learning algorithms without any prior background start showing better perfomance.
Deep Learning is a promising technology and computational resources are available to run such models. However, big annotated datasets are required to get good accuracy from this algorithms.
The POINTOUT project is aimed at:
Developing a platform to build and share collaborative training datasets for combined EO/ML communities
Implementing generic DL algorithms to detect objects in EO scenes for expert and/or non-expert users online
Developed by:
Funded by:
