A realistic fish-habitat dataset to evaluate algorithms for underwater visual analysis

Saleh, Alzayat, Laradji, Issam H., Konovalov, Dmitry A., Bradley, Michael, Vazquez, David, and Sheaves, Marcus (2020) A realistic fish-habitat dataset to evaluate algorithms for underwater visual analysis. Scientific Reports, 10. 14671.

PDF (Published Version) - Published Version
Available under License Creative Commons Attribution.

Download (1MB) | Preview
View at Publisher Website: https://doi.org/10.1038/s41598-020-71639...


Visual analysis of complex fish habitats is an important step towards sustainable fisheries for human consumption and environmental protection. Deep Learning methods have shown great promise for scene analysis when trained on large-scale datasets. However, current datasets for fish analysis tend to focus on the classification task within constrained, plain environments which do not capture the complexity of underwater fish habitats. To address this limitation, we present DeepFish as a benchmark suite with a large-scale dataset to train and test methods for several computer vision tasks. The dataset consists of approximately 40 thousand images collected underwater from 20 habitats in the marine-environments of tropical Australia. The dataset originally contained only classification labels. Thus, we collected point-level and segmentation labels to have a more comprehensive fish analysis benchmark. These labels enable models to learn to automatically monitor fish count, identify their locations, and estimate their sizes. Our experiments provide an in-depth analysis of the dataset characteristics, and the performance evaluation of several state-of-the-art approaches based on our benchmark. Although models pre-trained on ImageNet have successfully performed on this benchmark, there is still room for improvement. Therefore, this benchmark serves as a testbed to motivate further development in this challenging domain of underwater computer vision.

Item ID: 64493
Item Type: Article (Research - C1)
ISSN: 2045-2322
Related URLs:
Copyright Information: © The Author(s) 2020.
Additional Information:

A version of this publication was included as Chapter 4 of the following PhD thesis: Saleh, Alzayat (2020) Developing deep learning methods for aquaculture applications. Masters (Research) thesis, James Cook University, which is available Open Access in ResearchOnline@JCU. Please see the Related URLs for access.

Funders: Australian Research Training Scholarship, James Cook University (JCU), UBC Four-Year Doctoral Fellowships
Projects and Grants: JCU Strategic Research Initiative Funding SRIF-2018
Date Deposited: 06 Oct 2020 21:19
FoR Codes: 46 INFORMATION AND COMPUTING SCIENCES > 4611 Machine learning > 461199 Machine learning not elsewhere classified @ 100%
SEO Codes: 96 ENVIRONMENT > 9605 Ecosystem Assessment and Management > 960507 Ecosystem Assessment and Management of Marine Environments @ 100%
Downloads: Total: 732
Last 12 Months: 123
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page