Abstract:Underwater imaging enables nondestructive plankton sampling at frequencies, durations, and resolutions unattainable by traditional methods. These systems necessitate automated processes to identify organisms efficiently. Early underwater image processing used a standard approach: binarizing images to segment targets, then integrating deep learning models for classification. While intuitive, this infrastructure has limitations in handling high concentrations of biotic and abiotic particles, rapid changes in dominant taxa, and highly variable target sizes. To address these challenges, we introduce a new framework that starts with a scene classifier to capture large within-image variation, such as disparities in the layout of particles and dominant taxa. After scene classification, scene-specific Mask regional convolutional neural network (Mask R-CNN) models are trained to separate target objects into different groups. The procedure allows information to be extracted from different image types, while minimizing potential bias for commonly occurring features. Using in situ coastal plankton images, we compared the scene-specific models to the Mask R-CNN model encompassing all scene categories as a single full model. Results showed that the scene-specific approach outperformed the full model by achieving a 20% accuracy improvement in complex noisy images. The full model yielded counts that were up to 78% lower than those enumerated by the scene-specific model for some small-sized plankton groups. We further tested the framework on images from a benthic video camera and an imaging sonar system with good results. The integration of scene classification, which groups similar images together, can improve the accuracy of detection and classification for complex marine biological images.

Multi-dimensional,multi-level Text Classification System for Marine Knowledge

Application of Text Classification in Marine Information Processing

Marine Information Gathering, Transmission, Processing, and \\fusion: Current Status and Future Trends

Multi-class Fish Stock Statistics Technology Based on Object Classification and Tracking Algorithm

Marine Information Technology: the Best is Yet to Come

Design and Implementation of Marine Multi-Source Heterogeneous Data Transform System

Research and Implementation of Multi-source Integrated Marine Data Management Application Platform

Design and Implementation of Marine Data Management System

Study on the Text Classification Algorithm Based on Knowledge Ontology

Design and Application of Marine Remote Sensing Spatial Database Engine for Web Publishing

A real-time publication system of massive marine remote sensing data - art. no. 67951V

The Design and Implement of Marine Remote Sensing Multidimensional Dynamic Visualization System Based on COM

Comprehensive multi-model shared method of marine dynamic monitoring data

Multi-label classification for simultaneous fault diagnosis of marine machinery: A comparative study

Text Classification Based on Natural Language Processing and Machine Learning in Multi Label Corpus

Multi Features Combination for Automated Zooplankton Classification

Application of COM Technique in Marine Remote Sensing Multidimensional Dynamic Visualization System

Implementation of Data Management and Analysis System for Marine Ranching

Design of Visual Query and Display for Marine Data

Application Study on Marine Data Mining Technology

Taming the Data Deluge: a Novel End-to-end Deep Learning System for Classifying Marine Biological and Environmental Images