Precise automatic classification of 46 different pollen types with convolutional neural networks

In palynology, the visual classification of pollen grains from different species is a hard task which is usually tackled by human operators using microscopes. Many industries, including medical and farmaceutical, rely on the accuracy of this manual classification process, which is reported to be around 67%. In this paper, we propose a new method to automatically classify pollen grains using deep learning techniques that improve the correct classification rates in images not previously seen by the models. Our proposal manages to properly classify up to 98% of the examples from a dataset with 46 different classes of pollen grains, produced by the Classifynder classification system. This is an unprecedented result which surpasses all previous attempts both in accuracy and number and difficulty of taxa under consideration, which include types previously considered as indistinguishable.

[1]  K. Holt,et al.  Principles and methods for automated palynology. , 2014, The New phytologist.

[2]  V. Bryant,et al.  Forensic palynology: why do it and how it works. , 2006, Forensic science international.

[3]  M. Bebbington,et al.  Separating Morphologically Similar Pollen Types Using Basic Shape Features from Digital Images: A Preliminary Study , 2014 .

[4]  Víctor Sevillano,et al.  Improving classification of pollen grain images of the POLEN23E dataset through three different applications of deep learning convolutional neural networks , 2018, PloS one.

[5]  K. Larsen.  Pollen grains of New Zealand dicotyledonous plants , 1994 .

[6]  K. Bastl,et al.  Evaluation of Pollen Apps Forecasts: The Need for Quality Control in an eHealth Service , 2017, Journal of medical Internet research.

[7]  L. Lourens,et al.  Rapid climate change from north Andean Lake Fúquene pollen records driven by obliquity: Implications for a basin-wide biostratigraphic zonation for the last 284 ka , 2011 .

[8]  R. Niessner,et al.  Characterization and discrimination of pollen by Raman microscopy , 2005, Analytical and bioanalytical chemistry.

[9]  R. M. Hodgson,et al.  Progress towards an automated trainable pollen location and classifier system for use in the palynology laboratory , 2011 .

[10]  Mark Bush,et al.  Pollen Grain Recognition Using Deep Learning , 2016, ISVC.

[11]  Ariadne Barbosa Gonçalves,et al.  Feature Extraction and Machine Learning for the Classification of Brazilian Savannah Pollen Grains , 2016, PloS one.

[12]  G. Erdtman.  The acetolysis method-a revised description , 1960 .

[13]  Bent Vad Odgaard,et al.  Fossil pollen as a record of past biodiversity , 1999 .

[14]  L. P. Oddo,et al.  Harmonized methods of melissopalynology , 2004 .

[15]  Abdelmalek Amine,et al.  Deep Convolutional Neural Network for Pollen Grains Classification , 2019, JERI.

[16]  V. Rull,et al.  Climate variability and human impact in South America during the last 2000 years: synthesis and perspectives from pollen records , 2016 .

[17]  Andrey Filchenkov,et al.  Pollen grain recognition using convolutional neural network , 2018, ESANN.

[18]  Mark Bush,et al.  Sequential Recognition of Pollen Grain Z-Stacks by Combining CNN and RNN , 2018, FLAIRS.

[19]  Yulia Arzhaeva,et al.  A comparison of classification algorithms within the Classifynder pollen imaging system , 2013 .

[20]  S. Bonini,et al.  Allergenic pollen and pollen allergy in Europe , 2007, Allergy.

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

[22]  R. Dell’Anna,et al.  Pollen discrimination and classification by Fourier transform infrared (FT-IR) microspectroscopy and machine learning , 2009, Analytical and bioanalytical chemistry.

[23]  Katsumi Yabusaki,et al.  Classification of pollen species using autofluorescence image analysis. , 2009, Journal of bioscience and bioengineering.