Abstract
This paper presents the evaluation of 36 convolutional neural network (CNN) models, which were trained on the same dataset (ImageNet). The aim of this research was to evaluate the performance of pre-trained models on the binary classification of images in a “real-world” application. The classification of wildlife images was the use case, in particular, those of the Eurasian lynx (lat. “Lynx lynx”), which were collected by camera traps in various locations in Croatia. The collected images varied greatly in terms of image quality, while the dataset itself was highly imbalanced in terms of the percentage of images that depicted lynxes.
Highlights
In the present article, the authors suggest the use of various convolutional neural network models as a tool to help scientists classify images according to their content
This paper presents the evaluation of 36 convolutional neural network (CNN) models, which were trained on the same dataset (ImageNet)
Based on the results of the classification process and the boolean flag whether the lynx was really depicted in the image, basic confusion matrix parameters true positive (TP), true negative (TN), false positive (FP), and false negative (FN) were calculated
Summary
The authors suggest the use of various convolutional neural network models as a tool to help scientists classify images according to their content. All classified images were collected as part of other projects that have studied animal behavior and migration in mountainous and wooded parts of Croatia, Gorski Kotar (https://en.wikipedia.org/wiki/Gorski_Kotar, accessed on 25 September 2021), Risnjak (https://en.wikipedia.org/wiki/Risnjak, accessed on 25 September 2021), and Lika (https://en.wikipedia.org/wiki/Lika, accessed on 25 September 2021). The initial purpose of this paper was to help our colleagues of the Wildlife and Environmental/Nature Protection Department at our institution speed up the analysis and classification of the massive number of camera trap images collected. One of the projects focused on the exploration of lynx behavior, habits, and migration and monitoring the number of individual animals in the population. The CNN models described in this paper have different architectures, numbers of parameters, and complexities, which influence their classification rate and accuracy
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
