Using Deep Learning to Identify Costa Rican Native Tree Species From Wood Cut Images

Figueroa-Mata, Geovanni; Mata‐Montero, Erick; Valverde, Juan Carlos; Arias‐Aguilar, Dagoberto; Zamora-Villalobos, Nelson

doi:10.3389/fpls.2022.789227

Cited by 10 publications

(5 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this study, we analyzed the classification performance and its influencing factors on the classification of softwood species in four architectures: VGG16, ResNet50, GoogLeNet, and basic CNN. The four architectures based on neural networks showed excellent classification performance of over 90% for wood species classification, showing similar performance as in previous studies [22][23][24][25]. In the architectures, the accuracy increased and the loss decreased with increasing epochs.…”

Section: Discussionsupporting

confidence: 82%

Performance Influencing Factors of Convolutional Neural Network Models for Classifying Certain Softwood Species

Kim,

Purusatama,

Savero

et al. 2023

Forests

View full text Add to dashboard Cite

This study aims to verify the wood classification performance of convolutional neural networks (CNNs), such as VGG16, ResNet50, GoogLeNet, and basic CNN architectures, and to investigate the factors affecting classification performance. A dataset from 10 softwood species consisted of 200 cross-sectional micrographs each from the total part, earlywood, and latewood of each species. We used 80% and 20% of each dataset for training and testing, respectively. To improve the performance of the architectures, the dataset was augmented, and the differences in classification performance before and after augmentation were compared. The four architectures showed a high classification accuracy of over 90% between species, and the accuracy increased with increasing epochs. However, the starting points of the accuracy, loss, and training speed increments differed according to the architecture. The latewood dataset showed the highest accuracy. The epochs and augmented datasets also positively affected accuracy, whereas the total part and non-augmented datasets had a negative effect on accuracy. Additionally, the augmented dataset tended to derive stable results and reached a convergence point earlier. In the present study, an augmented latewood dataset was the most important factor affecting classification performance and should be used for training CNNs.

show abstract

Section: Discussionsupporting

confidence: 82%

Performance Influencing Factors of Convolutional Neural Network Models for Classifying Certain Softwood Species

Kim,

Purusatama,

Savero

et al. 2023

Forests

View full text Add to dashboard Cite

show abstract

“…Efforts to standardize the image format of digitized specimens [35] has improved their value for morphological studies. The use of digitized herbarium specimens in informing AI-driven models for plant identification, morphological, and phenological studies is of particular note [36][37][38][39][40], but having a better idea of the taxonomic error rates in varying groups of plants before using these models will be of utmost importance as a model informed by misidentified specimens could perpetuate bias and incorrect information [24]. Additionally, previous studies have assessed herbarium images by themselves on a large scale and found some complexity regarding taxonomic identification and accuracy, largely cases of outdated taxonomy, but in some cases misspellings or misidentifications [41].…”

Section: Museum Collection Data and Herbarium Digitizationmentioning

confidence: 99%

Quantifying error in occurrence data: Comparing the data quality of iNaturalist and digitized herbarium specimen data in flowering plant families of the southeastern United States

White,

Soltis,

Soltis

et al. 2023

PLoS ONE

View full text Add to dashboard Cite

iNaturalist has the potential to be an extremely rich source of organismal occurrence data. Launched in 2008, it now contains over 150 million uploaded observations as of May 2023. Based on the findings of a limited number of past studies assessing the taxonomic accuracy of participatory science-driven sources of occurrence data such as iNaturalist, there has been concern that some portion of these records might be misidentified in certain taxonomic groups. In this case study, we compare Research Grade iNaturalist observations with digitized herbarium specimens, both of which are currently available for combined download from large data aggregators and are therefore the primary sources of occurrence data for large-scale biodiversity/biogeography studies. Our comparisons were confined regionally to the southeastern United States (Florida, Georgia, North Carolina, South Carolina, Texas, Tennessee, Kentucky, and Virginia). Occurrence records from ten plant families (Gentianaceae, Ericaceae, Melanthiaceae, Ulmaceae, Fabaceae, Asteraceae, Fagaceae, Cyperaceae, Juglandaceae, Apocynaceae) were downloaded and scored on taxonomic accuracy. We found a comparable and relatively low rate of misidentification among both digitized herbarium specimens and Research Grade iNaturalist observations within the study area. This finding illustrates the utility and high quality of iNaturalist data for future research in the region, but also points to key differences between data types, giving each a respective advantage, depending on applications of the data.

show abstract

“…Recent research studies have demonstrated the high accuracy rates achieved by these models in the classification of flora images ( Kattenborn et al., 2020 ). For instance, a study focusing on the classification of Brazilian flora images using deep learning models reported classification accuracies reaching up to 93% for InceptionV3 and ResNet152 ( Figueroa-Mata et al., 2022 ).…”

Section: Related Workmentioning

confidence: 99%

Deep learning for Chilean native flora classification: a comparative analysis

Figueroa-Flores,

San-Martin

2023

Front. Plant Sci.

View full text Add to dashboard Cite

The limited availability of information on Chilean native flora has resulted in a lack of knowledge among the general public, and the classification of these plants poses challenges without extensive expertise. This study evaluates the performance of several Deep Learning (DL) models, namely InceptionV3, VGG19, ResNet152, and MobileNetV2, in classifying images representing Chilean native flora. The models are pre-trained on Imagenet. A dataset containing 500 images for each of the 10 classes of native flowers in Chile was curated, resulting in a total of 5000 images. The DL models were applied to this dataset, and their performance was compared based on accuracy and other relevant metrics. The findings highlight the potential of DL models to accurately classify images of Chilean native flora. The results contribute to enhancing the understanding of these plant species and fostering awareness among the general public. Further improvements and applications of DL in ecology and biodiversity research are discussed.

show abstract

Using Deep Learning to Identify Costa Rican Native Tree Species From Wood Cut Images

Cited by 10 publications

References 23 publications

Performance Influencing Factors of Convolutional Neural Network Models for Classifying Certain Softwood Species

Performance Influencing Factors of Convolutional Neural Network Models for Classifying Certain Softwood Species

Quantifying error in occurrence data: Comparing the data quality of iNaturalist and digitized herbarium specimen data in flowering plant families of the southeastern United States

Deep learning for Chilean native flora classification: a comparative analysis

Contact Info

Product

Resources

About