Ant genera identification using an ensemble of convolutional neural networks

Marques, Alan Caio R.; Raimundo, Marcos M.; Cavalheiro, Ellen Marianne Bernal; Salles, Luis; Filho, Christiano Lyra; Zuben, Fernando J. Von

doi:10.1371/journal.pone.0192011

Cited by 34 publications

(39 citation statements)

References 21 publications

(17 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Image classifications used for species identification have dramatically increased in accuracy, performance, and in the number of taxa analyzed (Marques et al, 2018;Martineau et al, 2017;Norouzzadeh et al, 2018;Schneider, Taylor, & Kremer, 2018;Van Horn et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

“…Comparing our results to those obtained from other convolutional neural networks, built for the specific purpose of identifying other groups of arthropods (e.g.,Marques et al, 2018), there is scope for increasing both classification recall and taxonomic resolution.Either image quality, network structure, number of classes to predict, or the image recording perspective could explain the differences.When comparing precision and accuracy of dorsal perspective images of ants byMarques et al (2018) we achieve comparable results (precision 54.7% and balanced accuracy 75.3% vs. 52.0% and 59.0%).For a range of other studies that classify arthropods to species level, the results are comparable, even though fewer species are typically used in classification(Martineau et al, 2017). van Horn et al(2017) presented a species level trained network based on Inception ResNet v2 and 675,000 images among 5,000 species of plants and animals.…”

mentioning

confidence: 99%

“…Explanatory variables included the top1 value in the output layer, body size in megapixels, number of species in the same genus as the ground truth species, number of training images of the ground truth species.Comparing our results to those obtained from other convolutional neural networks, built for the specific purpose of identifying other groups of arthropods (e.g.,Marques et al, 2018), there is scope for increasing both classification recall and taxonomic resolution.Either image quality, network structure, number of classes to predict, or the image recording perspective could explain the differences.When comparing precision and accuracy of dorsal perspective images of ants byMarques et al (2018) we achieve comparable results (precision 54.7% and balanced accuracy 75.3% vs. 52.0% and 59.0%).…”

mentioning

confidence: 99%

See 2 more Smart Citations

Species‐level image classification with convolutional neural network enables insect identification from habitus images

Hansen

Svenning

Olsen

et al. 2019

Ecology and Evolution

View full text Add to dashboard Cite

1. Changes in insect biomass, abundance, and diversity are challenging to track at sufficient spatial, temporal, and taxonomic resolution. Camera traps can capture habitus images of ground-dwelling insects. However, currently sampling involves manually detecting and identifying specimens. Here, we test whether a convolutional neural network (CNN) can classify habitus images of ground beetles to species level, and estimate how correct classification relates to body size, number of species inside genera, and species identity.2. We created an image database of 65,841 museum specimens comprising 361 carabid beetle species from the British Isles and fine-tuned the parameters of a pretrained CNN from a training dataset. By summing up class confidence values within genus, tribe, and subfamily and setting a confidence threshold, we trade-off between classification accuracy, precision, and recall and taxonomic resolution.3. The CNN classified 51.9% of 19,164 test images correctly to species level and 74.9% to genus level. Average classification recall on species level was 50.7%.Applying a threshold of 0.5 increased the average classification recall to 74.6% at the expense of taxonomic resolution. Higher top value from the output layer and larger sized species were more often classified correctly, as were images of species in genera with few species. 4. Fine-tuning enabled us to classify images with a high mean recall for the whole test dataset to species or higher taxonomic levels, however, with high variability.This indicates that some species are more difficult to identify because of properties such as their body size or the number of related species.5. Together, species-level image classification of arthropods from museum collections and ecological monitoring can substantially increase the amount of occurrence data that can feasibly be collected. These tools thus provide new opportunities in understanding and predicting ecological responses to environmental change. Jens-Christian Svenninghttps://orcid.

show abstract

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Species‐level image classification with convolutional neural network enables insect identification from habitus images

Hansen

Svenning

Olsen

et al. 2019

Ecology and Evolution

View full text Add to dashboard Cite

show abstract

“…leaves, fruits, flower for plants) or perspectives (e.g. head, dorsum and profile for insects) (Lee et al., ; Marques et al., ). Analogous to a biologist that generally tries identifying an organism by observing several organs or a similar organ from different viewpoints, an important research direction is analyzing how to increase accuracy by combining different perspectives in an automated identification.…”

Section: Discussionmentioning

confidence: 99%

Machine learning for image based species identification

Wäldchen¹,

2018

View full text Add to dashboard Cite

Accurate species identification is the basis for all aspects of taxonomic research and is an essential component of workflows in biological research. Biologists are asking for more efficient methods to meet the identification demand. Smart mobile devices, digital cameras as well as the mass digitisation of natural history collections led to an explosion of openly available image data depicting living organisms. This rapid increase in biological image data in combination with modern machine learning methods, such as deep learning, offers tremendous opportunities for automated species identification. In this paper, we focus on deep learning neural networks as a technology that enabled breakthroughs in automated species identification in the last 2 years. In order to stimulate more work in this direction, we provide a brief overview of machine learning frameworks applicable to the species identification problem. We review selected deep learning approaches for image based species identification and introduce publicly available applications. Eventually, this article aims to provide insights into the current state‐of‐the‐art in automated identification and to serve as a starting point for researchers willing to apply novel machine learning techniques in their biological studies. While modern machine learning approaches only slowly pave their way into the field of species identification, we argue that we are going to see a proliferation of these techniques being applied to the problem in the future. Artificial intelligence systems will provide alternative tools for taxonomic identification in the near future.

show abstract

“…CNNs have been applied in a wide range of applications from face recognition to identification of actual neurons in microscopic images [19,20]. In the ecological domain, CNNs have been applied in diverse settings such as detection of insects, wildlife in terrestrial ecosystems, and scallops on the sea floor [21][22][23].…”

Section: Introductionmentioning

confidence: 99%

Automated classification of three-dimensional reconstructions of coral reefs using convolutional neural networks

et al. 2020

View full text Add to dashboard Cite

Coral reefs are biologically diverse and structurally complex ecosystems, which have been severally affected by human actions. Consequently, there is a need for rapid ecological assessment of coral reefs, but current approaches require time consuming manual analysis, either during a dive survey or on images collected during a survey. Reef structural complexity is essential for ecological function but is challenging to measure and often relegated to simple metrics such as rugosity. Recent advances in computer vision and machine learning offer the potential to alleviate some of these limitations. We developed an approach to automatically classify 3D reconstructions of reef sections and assessed the accuracy of this approach. 3D reconstructions of reef sections were generated using commercial Structurefrom-Motion software with images extracted from video surveys. To generate a 3D classified map, locations on the 3D reconstruction were mapped back into the original images to extract multiple views of the location. Several approaches were tested to merge information from multiple views of a point into a single classification, all of which used convolutional neural networks to classify or extract features from the images, but differ in the strategy employed for merging information. Approaches to merging information entailed voting, probability averaging, and a learned neural-network layer. All approaches performed similarly achieving overall classification accuracies of~96% and >90% accuracy on most classes. With this high classification accuracy, these approaches are suitable for many ecological applications.

show abstract

Ant genera identification using an ensemble of convolutional neural networks

Cited by 34 publications

References 21 publications

Species‐level image classification with convolutional neural network enables insect identification from habitus images

Species‐level image classification with convolutional neural network enables insect identification from habitus images

Machine learning for image based species identification

Automated classification of three-dimensional reconstructions of coral reefs using convolutional neural networks

Contact Info

Product

Resources

About