Abstract:Anthropogenic actions have caused severe environmental changes, leading to a rapid and massive loss of biodiversity, called the Holocene Extinction or "Sixth Extinction." In this context, integrative methods, capable of quickly and reliably determining and differentiating species, have become constantly demanded in taxonomic studies. Near-infrared spectroscopical technologies have proved to be promising for application in the integrative taxonomy of insects, and it is a true-non-destructive method, causing no damage to the sample and being completely chemical-preparation-free. Near-infrared spectral profiles are known as the fingerprints of the chemical composition of a given sample, and a new layer of information may be accessed. Hyperspectral imaging technologies in the near-infrared range are among the most popular, although their usage is still incipient in insect studies. As in other animal taxa, katydids' taxonomy has traditionally been based on morphological comparison, resulting in many misclassifications over the years. However, integrative methods are more and more required in taxonomic studies. Different methods and technologies have been used from an integrative perspective to minimize misidentifications, especially for non-taxonomist or untrained researchers. Here, we approach the applicability of near-infrared spectroscopy coupled with hyperspectral imaging technology to Ensifera specimens housed in collections, discussing the advantages, disadvantages, and challenges to future applications. As a result, we propose using only homologous body parts in comparing and modeling species using this kind of data due to the heterogeneity of the insects' exoskeleton. Additionally, we made a case study by discriminating four species of the katydid genus Conocephalus Thunberg, one of the most speciose genera within Orthoptera that is known to have polymorphic species, with variations expressed in the wing development and postabdomen appendages. We generated a Partial Least Squares-Discriminant Analysis (PLS-DA) classification model for the species with an overall classification accuracy for the assigned pixels of 90% and specific accuracy for pixel discrimination ranging from 96% to 98%. This is one of the few research studies employing hyperspectral imaging in insects' taxonomy, and it is the very first to use this technology in Orthoptera; therefore, this is a preliminary approach to usage as an integrative method.

Developing a Computer Application for the Identification of Similar Butterfly Species Using MATLAB Image Processing

A Computational Approach for Recognition of Electronic Microscope Plant Pollen Images

Towards the Automated Identification of Chrysomya Blow Flies from Wing Images

Pest Identification using Image Processing Techniques in Detecting Image Pattern through Neural Network

Deep learning powered real-time identification of insects using citizen science data

AInsectID Version 1.1: an Insect Species Identification Software Based on the Transfer Learning of Deep Convolutional Neural Networks

Paddy insect identification using deep features with lion optimization algorithm

Plant species identification using color learning resources, shape, texture, through machine learning and artificial neural networks

Morphology-Based Identification of Bemisia tabaci Cryptic Species Puparia via Embedded Group-Contrast Convolution Neural Network Analysis

Artificial intelligence correctly classifies developmental stages of monarch caterpillars enabling better conservation through the use of community science photographs

Study On Field Weed Recognition In Real-Time

Spotted! Computer-aided individual photo-identification allows for mark-recapture of invasive spotted lanternfly (Lycorma delicatula)

Insect Pest Image Recognition: A Few-Shot Machine Learning Approach including Maturity Stages Classification

Wing Interferential Patterns (WIPs) and machine learning for the classification of some Aedes species of medical interest

Images in red: A methodological and integrative approach for the usage of Near-infrared Hyperspectral Imaging (NIR-HSI) on collection specimens of Orthoptera (Insecta)

3CCD Image Segmentation and Edge Detection Based on MATLAB

Insect Identification in the Wild: The AMI Dataset

Extraction of Pest Insect Characteristics Present in a Mirasol Pepper (Capsicum annuum L.) Crop by Digital Image Processing

Machine learning ensemble with image processing for pest identification and classification in field crops

Automated identification of aquatic insects: A case study using deep learning and computer vision techniques

Machine Learning Challenges of Biological Factors in Insect Image Data