Figure 3: The data workflow as applied to the AUV use case. | Scientific Data

Figure 3: The data workflow as applied to the AUV use case.

From: An acquisition, curation and management workflow for sustainable, terabyte-scale marine image analysis

Figure 3

The AUV Abyss created metadata files (a) and stacks of up to 50,000 images (b) per dive. Meta- and image data were fused by time code (d). Un-distortion was applied (e), erroneous data were removed (f). Raw metadata are stored in OSIS (c). Raw and curated imagery is managed with ProxSys (g, h). Curated image data are made publicly available: in PANGAEA for long-term archival (i - by duplication) and in BIIGLE for manual annotation (j - by link). OSIS links to the image data in BIIGLE (k). Subsequent image analysis, enabled by the curated data are color normalization (i), mosaicking (n), mineral resource mapping (o), and automated event detection and classification within individual images (p), using manual annotations from BIIGLE (m) and machine learning.

Back to article page