PSS-net: Parallel semantic segmentation network for detecting marine animals in underwater scene

Abstract

Marine scene segmentation is a core technology in marine biology and autonomous underwater vehicle research. However, it is challenging from the perspective of having a different environment from that of the conventional traffic segmentation on roads. There are two major challenges. The first is the difficulty of searching for objects under seawater caused by the relatively low-light environment. The second problem is segmenting marine animals with protective colors. To solve such challenges, in previous research, a method of simultaneously segmenting the foreground and the background was proposed based on a simple modification of the conventional model; however, it has limitations in improving the segmentation accuracy. Therefore, we propose a parallel semantic segmentation network to solve the above issues in which a model and a loss are employed to locate the foreground and the background separately. The training task to locate the foreground and the background is reinforced in the proposed method by adding an attention technique in a parallel model. Furthermore, the final segmentation is performed by aggregating two feature maps obtained by separately locating the foreground and the background.The test results using an open dataset for marine animal segmentation reveal that the proposed method achieves performance of 87%, 97.3%, 88%, 95.2%, and 0.029 in the mean intersection of the union, structure similarities, weighted F-measure, enhanced-alignment measure, and mean absolute error, respectively. These findings confirm that the proposed method has higher accuracy than the state-of-the-art methods. The proposed model and code are publicly available via Github1.

A Deep-Learning Based Pipeline for Estimating the Abundance and Size of Aquatic Organisms in an Unconstrained Underwater Environment from Continuously Captured Stereo Video

“…This type of segmentation is particularly useful for a more accurate estimation of animal sizes, as it provides the exact image area occupied by the animal. Various deep learning models have been applied successfully for free-swimming animals, such as cod and jellyfish, captured by stationary underwater cameras [ 46 ], up to 37 different marine animals, including those with camouflage, as part of the MAS3K dataset [ 47 , 48 , 49 ], fish passing through a camera chamber in a fish trawl [ 40 ] and multiple fish species in a tropical habitat [ 50 ].…”

Underwater Organism Color Fine-Tuning via Decomposition and Guidance

“…Under this setting, the color code will be learned implicitly. CECF can adjust the color of a specific organism when the mask of this organism is provided by well-studied segmentation algorithms Nezla, Haridas, and Supriya 2021;Kim and Park 2022;Alshdaifat, Talib, and Osman 2020). Since segmentation is not the research scope of this paper, we do not discuss how to segment underwater organisms.…”

MSGNet: multi-source guidance network for fish segmentation in underwater videos

“…Natural underwater environments, often with complex backgrounds such as seagrass and reefs, interfere with foreground object localization. To improve the performance of fish segmentation in complex underwater scenes, Kim et al (Kim and Park, 2022). proposed a parallel semantic segmentation network that utilizes model and loss to localize the foreground and background, respectively and achieves efficient detection of marine animals by learning their foreground and background regions separately.…”