Research on Spider Sex Recognition From Images Based on Deep Learning

Abstract

The rapid and accurate identification of spider sex is the first step in spider image recognition. The traditional artificial method used to identify the sex of mature spiders is mainly based on their genital structures (male palps or female epigynum) and highly dependent on the professional background of the identifiers. This article uses computer-based deep learning and transfer learning to identify the sex of spider, explores the design and application of convolutional neural networks in deep learning for spider sex recognition from images, and establishes a neural network model that displays excellent performance in experiments. In addition to optimizing the network model, we select appropriate hyperparameters to improve the accuracy of recognition and reduce the influence of human factors in the identification process. Through a comparison of multiple sets of experiments based on existing sample data collected in the laboratory, we find that the transfer learning method based on Xception can obtain better prediction accuracy than ResNet-152. After data augmentation, the optimization of a combined activation function and the fine-tuning of frozen layers, the prediction accuracy reaches 98.02%, and for an actual measurement of independent samples, the recognition accuracy reaches 92.38%. Therefore, the proposed method can basically replace manual identification and provide a reference for the artificial intelligence-based identification of spider species. Additionally, the model results indicate that male and female dimorphism may exist beyond the non-genital characteristics of spiders. INDEX TERMS deep learning; transfer learning; convolutional neural network; spider sex identification

SpiderID_APP: A User-Friendly APP for Spider Identification in Taiwan Using YOLO-Based Deep Learning Models

“…In the realm of academic research, two other studies have focused on spider identification, considering taxonomy and gender [27,28]. Unfortunately, comparisons with these studies are not feasible.…”

“…Unfortunately, comparisons with these studies are not feasible. One study did not release its model or application [27], while the other claimed to have developed an iOS application [28]. However, their associated GitHub repository is outdated, making a comparative test impossible.…”

“…The field of automatic classification is rapidly growing with the application of deep learning and artificial intelligence [25,26]. Mainly, image classification for spider identification using a deep learning approach still needs improvement compared to other fields [27,28]. It is observed that accuracy evaluation on testing datasets in such studies is high due to the low diversity in the dataset used.…”

“…It is observed that accuracy evaluation on testing datasets in such studies is high due to the low diversity in the dataset used. For example, a study conducted by Sinnott et al in 2020 used nine classification classes belonging to nine different families within two infraorders [28], while the research by Chen et al in 2021 used multi-angle images of 30 species within a single genus of Pseudopoda for gender classification [27].…”

Spider Species Classification Using Vision Transformers and Convolutional Neural Networks

“…They report F1-score that ranges from 0.85 to 0.94. In [9], authors address the problem of spider sex recognition. Their dataset comprises 3,133 exemplars and report an accuracy of 92,38% on the validation set.…”