Australian sea-floor survey data, with images and expert annotations

Area of Science:

• Marine Biology
• Oceanography
• Robotics

Background:

• Benthic habitat mapping is crucial for marine ecosystem understanding.
• Autonomous Underwater Vehicles (AUVs) offer efficient data collection for large-scale surveys.
• Expert annotation of imagery is time-consuming but essential for accurate habitat classification.

Purpose of the Study:

• To create a comprehensive, expert-annotated benthic image dataset for Australian waters.
• To provide georeferenced data for studying marine habitats and organisms.
• To facilitate advancements in marine science, robotics, and computer vision.

Main Methods:

• Data collected using an Autonomous Underwater Vehicle (AUV) around Australia.
• Expert labeling of 407,968 points using the CATAMI classification scheme.
• Integration of data from multiple Australian research groups and the Integrated Marine Observing System (IMOS).

Main Results:

• A large-scale, georeferenced dataset of benthic images with expert annotations.
• Detailed information on substratum type and biota for coastal Australia.
• Associated sensor data, including geolocation, enhancing data utility.

Conclusions:

• The BENTHOZ-2015 dataset is a significant resource for marine science research.
• The data supports ecological studies, AUV development, and computer vision applications.
• This dataset contributes to Australia's ongoing marine biodiversity monitoring efforts.