Autonomous Plankton Classification from Reconstructed Holographic Imagery by L1- PCA-assisted Convolutional Neural Networks
Abstract
Studying and monitoring plankton distribution is vital for global climate and environment protection as they are the most elementary part of oceanic eco-systems. However, the conventional methods and techniques used for understanding the planktons are slow and lacks precision and therefore, in modern day scientific and engineering implementations, Convolutional Neural Networks is extensively used in deep learning and machine learning applications as it outperforms traditional manual approach. Dynamic nature of oceans make it very challenging to monitor these microscopic organisms. Our approach here is to generate a powerful automated plankton recognition system to autonomously identify them and improve the D-CNN for classification of the Plankton holographic imagery curated with the method of Data Conformity Evaluation. The performance of D-CNN classifier is improved by various hyper-parameter tuning, regularization techniques and appending meta-data. Conformity evaluation is based on a matric that's calculated on a continuously refined sequence of calculated $L_{1}$-norm tensor subspaces of the Plankton images. We note that our classifier performs accurately where our results improve performances from contemporary Deep Learning classifier alone.
BibTeX
@inproceedings{Vargas2024Contributing,
author = {Vargas, Luis Miguel Martínez and Laurido, Ana Lucía Caicedo and Latorre, Claudia Patricia Urbano and Correa, Yady Tatiana Solano and Ordóñez, Julián Fernando Muñoz},
booktitle = {2024 XVIII National Meeting on Optics and the IX Andean and Caribbean Conference on Optics and its Applications (ENO-CANCOA)},
title = {Contributing to Fishery Productivity in Colombia: A Machine Learning Approach to Predict Missing Chlorophyll-A Values Using MODIS Satellite Imagery},
year = {2024},
volume = {},
number = {},
pages = {1-6},
abstract = {The fisheries sector in Colombia plays a significant role, contributing 0.3% to the country's Gross Domestic Product (GDP) and generating exports worth US∃45.1 million, equivalent to 3.3% of the agricultural GDP. However, its management faces challenges such as non-target species fishing, inadequate control of overfishing, and resource management issues, among others, affecting fish production. This article highlights the necessity of enhancing chlorophyll-a measurement to op-timize fishery production. Chlorophyll-a measurements are vital indicators of marine ecosystem health. Utilizing satellite imagery like MODIS is crucial for accurate data collection. However, Colombia's geographic location, characterized by high cloud cover, compromises image quality for much of the year, posing significant limitations on reporting chlorophyll-a values. We propose a machine learning algorithm to predict chlorophyll-A values on MODIS images to address this issue. The approach demonstrates an accuracy exceeding 0.8 regarding R-squared for predicting missing chlorophyll-a values. By overcoming spatial limitations caused by cloud cover, this method enables a more precise assessment of fishing grounds. Various machine learning models were also applied and evaluated within the research's context. Results yielded a 5% recovery yield of chlorophyll-a values for 2023, enriching knowledge and management practices within Colombia's fishing sector.},
keywords = {Machine learning algorithms;Economic indicators;Biological system modeling;Clouds;Predictive models;Prediction algorithms;Fisheries;Optics;Satellite images;MODIS;Chlorophyll-a;Predict Values;MODIS Images;Machine Learning Models;Fishery Production;Cloud Cover},
doi = {10.1109/ENO-CANCOA61307.2024.10751189},
issn = {},
month = {June},
}