Deep learning in deep time

White, Alexander E.

doi:10.1073/pnas.2020870117

Cited by 2 publications

(1 citation statement)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Artificial intelligence (AI) can successfully identify species from images of live plants (Kumar et al, 2012; Joly et al, 2016; Tcheng et al, 2016; Rzanny et al, 2019; Champ et al, 2020; Minowa and Nagasaki, 2020; Unger et al, 2020) and herbarium sheets (Belhumeur et al, 2008; Unger et al, 2016; Carranza‐Rojas et al, 2017; Little et al, 2020; Romero et al, 2020). Machine learning identification of extant and fossil pollen at the species level has advanced significantly (Punyasena et al, 2012; Tcheng et al, 2016; Romero et al, 2020; White, 2020). Automated species identification of leaf images, in particular, is a well‐studied problem in computer vision (Im et al, 1998; Wu et al, 2007; Nam et al, 2008; Park et al, 2008; Caballero and Aranda, 2010; Bama et al, 2011; Hu et al, 2012; Laga et al, 2012; Larese et al, 2012; Mouine et al, 2012; Priya et al, 2012; Charters et al, 2014; Larese et al, 2014a, b; Jamil et al, 2015; Mata‐Montero and Carranza‐Rojas, 2015, 2016; Zhao et al, 2015; Grinblat et al, 2016; Larese and Granitto, 2016; Carranza‐Rojas, Mata‐Montero et al, 2018; Wäldchen and Mäder, 2018; Wäldchen et al, 2018; Almeida et al, 2020; Banerjee and Pamula, 2020; Bryson et al, 2020; Pryer et al, 2020; Soltis et al, 2020; Mukherjee et al, 2021; Zhou et al, 2021).…”

Section: Figurementioning

confidence: 99%

Decoding family‐level features for modern and fossil leaves from computer‐vision heat maps

Spagnuolo

Wilf

Serre

2022

American J of Botany

View full text Add to dashboard Cite

Premise Angiosperm leaves present a classic identification problem due to their morphological complexity. Computer‐vision algorithms can identify diagnostic regions in images, and heat map outputs illustrate those regions for identification, providing novel insights through visual feedback. We investigate the potential of analyzing leaf heat maps to reveal novel, human‐friendly botanical information with applications for extant‐ and fossil‐leaf identification. Methods We developed a manual scoring system for hotspot locations on published computer‐vision heat maps of cleared leaves that showed diagnostic regions for family identification. Heat maps of 3114 cleared leaves of 930 genera in 14 angiosperm families were analyzed. The top‐5 and top‐1 hotspot regions of highest diagnostic value were scored for 21 leaf locations. The resulting data were viewed using box plots and analyzed using cluster and principal component analyses. We manually identified similar features in fossil leaves to informally demonstrate potential fossil applications. Results The method successfully mapped machine strategy using standard botanical language, and distinctive patterns emerged for each family. Hotspots were concentrated on secondary veins (Salicaceae, Myrtaceae, Anacardiaceae), tooth apices (Betulaceae, Rosaceae), and on the little‐studied margins of untoothed leaves (Rubiaceae, Annonaceae, Ericaceae). Similar features drove the results from multivariate analyses. The results echo many traditional observations, while also showing that most diagnostic leaf features remain undescribed. Conclusions Machine‐derived heat maps that initially appear to be dominated by noise can be translated into human‐interpretable knowledge, highlighting paths forward for botanists and paleobotanists to discover new diagnostic botanical characters.

show abstract

Section: Figurementioning

confidence: 99%

Decoding family‐level features for modern and fossil leaves from computer‐vision heat maps

Spagnuolo

Wilf

Serre

2022

American J of Botany

View full text Add to dashboard Cite

show abstract

Application of neural networks for chitinozoans recognition in images

Yakupov,

Gusmanova

2024

Vestnik of Geosciences

View full text Add to dashboard Cite

Paleontological identification of microfauna using automated image recognition represents an innovative practical application of existing software methods for analysis and classification based on computer vision and machine learning technologies. The development of software capable of recognition of Chitinozoans in images will simplify and speed up the processing of large amounts of microfossil data. The use of neural networks for image analysis is also possible for other groups of paleoorganics. Chitinozoans have a number of advantages that allow a step-by-step assessment of the applicability of automated image recognition technology for biostratigraphic problems compared to other groups of microfossils. The artificial paleontological classification of Chitinozoa is based on clear morphological characteristics and can be formalized. At the first stage of solving recognition problems, a classification function was constructed that predicts class, the fossil belongs to, based on the input feature vector — either "chitinozoa" or "non-chitinozoa". The developed model of the Chitinozoans recognition algorithm showed a high degree of accuracy (more than 98 %).

show abstract

Deep learning in deep time

Cited by 2 publications

References 21 publications

Decoding family‐level features for modern and fossil leaves from computer‐vision heat maps

Decoding family‐level features for modern and fossil leaves from computer‐vision heat maps

Application of neural networks for chitinozoans recognition in images

Contact Info

Product

Resources

About