ALMI—A Generic Active Learning System for Computational Object Classification in Marine Observation Images

Abstract

In recent years, an increasing number of cabled Fixed Underwater Observatories (FUOs) have been deployed, many of them equipped with digital cameras recording high-resolution digital image time series for a given period. The manual extraction of quantitative information from these data regarding resident species is necessary to link the image time series information to data from other sensors but requires computational support to overcome the bottleneck problem in manual analysis. As a priori knowledge about the objects of interest in the images is almost never available, computational methods are required that are not dependent on the posterior availability of a large training data set of annotated images. In this paper, we propose a new strategy for collecting and using training data for machine learning-based observatory image interpretation much more efficiently. The method combines the training efficiency of a special active learning procedure with the advantages of deep learning feature representations. The method is tested on two highly disparate data sets. In our experiments, we can show that the proposed method ALMI achieves on one data set a classification accuracy A > 90% with less than N = 258 data samples and A > 80% after N = 150 iterations, i.e., training samples, on the other data set outperforming the reference method regarding accuracy and training data required.

Long‐term automated visual monitoring of Antarctic benthic fauna

“…Within this context, high‐definition still and video image data constitute a key non‐destructive approach for the non‐invasive monitoring of aquatic marine environments (Aguzzi et al, 2019; Bicknell et al, 2016; Jahanbakht et al, 2021). Many studies in the literature contributed to the analysis of benthic fauna outside the Antarctic region (Lopez‐Vazquez et al, 2020; Möller & Nattkemper, 2021), including the analysis of corals (Osterloff et al, 2019; Zuazo et al, 2020) and sponge dynamics (Harrison et al, 2021; Möller et al, 2019).…”

Deep learning–assisted biodiversity assessment in deep-sea benthic megafauna communities: a case study in the context of polymetallic nodule mining

“…In the case of benthic biodiversity community assessment, as considered here, the amount of image data collected often greatly exceeds the capacity of visual inspection by domain experts and the potential of machine learning-based annotation has gained interest over the last decade (e.g. Schoening et al, 2012;Möller and Nattkemper, 2021;Mbani et al, 2023;Yamada et al, 2023). To respond to the growing amounts of imagery, the BIIGLE system has been equipped with the MAIA tool (Machine learning Assisted Image Annotation) (Zurowietz et al, 2018;Zurowietz and Nattkemper, 2020) and cloud storage so users can upload their image data into the BIIGLE cloud, and analyse their image data more efficiently, employing machine learning.…”