Diatom Classification Including Morphological Adaptations Using CNNs

Diatoms are a major group of aquatic microalgae. They are widely used in different fields such as environmental studies to estimate water quality. This paper presents the use of convolutional neural networks (CNNs) to identify diatoms during their life cycle. This life cycle involves morphological and other changes to the diatom frustule adding intraclass variance and making harder the classification task. The performance of CNNs is compared against a classical image classification scheme (i.e., feature extraction and classification) using a 14 classes dataset with a total number of 1085 images ranging from 40 to 120 images per class. Classification accuracy was 99.07% and 99.7% for CNNs and classical methods respectively.

[1]  Janice L. Pappas,et al.  Legendre shape descriptors and shape group determination of specimens in the Cymbella cistula species complex , 2003 .

[2]  Gabriel Cristóbal,et al.  Diatom identification including life cycle stages through morphological and texture descriptors , 2019, PeerJ.

[3]  María Borrego-Ramos,et al.  Disentangling diatom species complexes: does morphometry suffice? , 2017, PeerJ.

[4]  David G. Mann,et al.  The Sellaphora pupula species complex (Bacillariophyceae): morphometric analysis, ultrastructure and mating data provide evidence for five new species , 2004 .

[5]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[6]  Tom M. Mitchell,et al.  Machine learning classifiers and fMRI: A tutorial overview , 2009, NeuroImage.

[7]  H. D. Buf,et al.  Automatic diatom identification , 2002 .

[8]  R. Fisher.  THE USE OF MULTIPLE MEASUREMENTS IN TAXONOMIC PROBLEMS , 1936 .

[9]  Geoffrey E. Hinton,et al.  Visualizing Data using t-SNE , 2008 .

[10]  Taghi M. Khoshgoftaar,et al.  A survey of transfer learning , 2016, Journal of Big Data.

[11]  James Bailey,et al.  Information Theoretic Measures for Clusterings Comparison: Variants, Properties, Normalization and Correction for Chance , 2010, J. Mach. Learn. Res..

[12]  Gabriel Cristóbal,et al.  Self-Invertible 2D Log-Gabor Wavelets , 2007, International Journal of Computer Vision.

[13]  Kilian Q. Weinberger,et al.  Densely Connected Convolutional Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[14]  Charles R. Giardina,et al.  Elliptic Fourier features of a closed contour , 1982, Comput. Graph. Image Process..

[15]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[16]  Hinrich Schütze,et al.  Introduction to information retrieval , 2008 .

[17]  Forrest N. Iandola,et al.  SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <1MB model size , 2016, ArXiv.

[18]  Gabriel Cristóbal,et al.  Automated Diatom Classification (Part A): Handcrafted Feature Approaches , 2017 .

[19]  Daiqing Mou,et al.  SEPARATING TABELLARIA (BACILLARIOPHYCEAE) SHAPE GROUPS BASED ON FOURIER DESCRIPTORS 1 , 1992 .

[20]  Peter Kovesi,et al.  Phase Congruency Detects Corners and Edges , 2003, DICTA.

[21]  Sergey Ioffe,et al.  Rethinking the Inception Architecture for Computer Vision , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[22]  Antanas Verikas,et al.  Phase congruency-based detection of circular objects applied to analysis of phytoplankton images , 2012, Pattern Recognit..

[23]  Gabriel Cristóbal,et al.  Automated Diatom Classification (Part B): A Deep Learning Approach , 2017 .