Skip to content

WP3: Data Analysis

Work Package Description

  • Work Package 3 will perform image analysis and implement a data workflow structure by building pipelines, as well as implement and evaluate data analysis. As such, this Work Package is the core of the SeaBee product generation engine. Work Package 3 will accommodate automated image analysis in a plug-and-play manner via advanced and cutting-edge Artificial Intelligence (AI) algorithms, as well as facilitate future improvements and developments within image and object analysis for identification of environmental and pollution relevant variables and respond to coming technologies.
  • Design & Methods: Drone data will be uploaded to the SeaBee data infrastructure and will be pre-processed and prepared for storage and analysis. The pre-processing approach will contain the following steps:
    • orthorectification,
    • image stitching,
    • radiometric calibration quality assessment of the data,
    • removal of personal/sensitive information and metadata.
  • Product generation will result from the use of state-of-the-art machine learning (ML) and other artificial intelligence (AI) protocols for object detection and thematic mapping of environmental data and creation of visualization products, e.g. marine habitat maps in coastal bays and inlets and automated recognition of marine mammals. Central for a successful implementation of state-of-the-art ML algorithms are use of graphics processing units (GPUs) and fast-working storage. This will be provided by the UNINETT/Sigma2 high performance computation infrastructure. A key part of the infrastructure is to implement a framework that enables users to teach an algorithm to perform a specific task, to create a desired product, or identify objects of interest. Further, protocols for re-analysis of previously recorded and stored data for new applications and/or research questions will be developed. This will provide powerful applications for reanalysis of previously collected data, for instance from seabird and mammal research, and for distribution of plastic debris. A protocol for preprocessing and data analysis will be developed, incl. description of the data format and level of preprocessing needed before uploading data to the SeaBee infrastructure (in Work Package 4). The preprocessing software will be built from available software, standard methods, and algorithms, and a part of Work Package 3 will be dedicated to access and acquire pre-processing software.

Lead: Norsk Regnesentral

Norsk Regnesentral (Norwegian Computing Center, NR) is a private, independent, non-profit foundation established in 1952. NR carries out contract research and development projects in the areas of information and communication technology and applied statistical modelling.


Arnt-Børre Salberg, NR. Arnt-Børre Salberg is a senior research scientist at NR and is specialised in Remote sensing, Machine learning, Deep learning, Image analysis, and Signal processing.