Megabenthic Fauna Detection with Faster R-CNN (FaunD-Fast) Short description of the research software.

Mbani, Benson , Buck, Valentin and Greinert, Jens (2022) Megabenthic Fauna Detection with Faster R-CNN (FaunD-Fast) Short description of the research software. Open Access DOI 10.3289/SW_1_2023.

[thumbnail of Archive copy of the software repository] Archive (Archive copy of the software repository)
faund-fast-main.zip - Submitted Version
Available under License Creative Commons: Attribution 4.0.

Download (20MB)

Supplementary data:

Abstract

This is an A.I. - based workflow for detecting megabenthic fauna from a sequence of underwater optical images. The workflow (semi) automatically generates weak annotations through the analysis of superpixels, and uses these (refined and semantically labeled) annotations to train a Faster R-CNN model. Currently, the workflow has been tested with images of the Clarion-Clipperton Zone in the Pacific Ocean

Document Type: Software
Keywords: Software; ARCHES; Digital twins; Digital Twin Prototype; Continuous Integration; Event-driven Architecture; ROS; Embedded Software Systems; Automated Testing
Research affiliation: OceanRep > GEOMAR > FB2 Marine Biogeochemistry > FB2-MG Marine Geosystems
Main POF Topic: PT6: Marine Life
Publisher: GEOMAR Helmholtz Centre for Ocean Research Kiel
Related URLs:
Date Deposited: 23 Mar 2023 09:43
Last Modified: 23 Mar 2023 10:29
URI: https://oceanrep.geomar.de/id/eprint/58250

Actions (login required)

View Item View Item