Enhanced Classification of Reptiles and Amphibians Using Deep Learning: A Comparative Analysis of Model Performance
Pratham Kaushik, Saniya Khurana · 2024
Reptiles and amphibians play a very imperative role in the maintenance of balance within ecosystems., given that they play dual roles: as predators and as prey., aside from being indicators of environmental stability. In this paper., researchers present a classification study using convolutional neural networks and transfer learning to identify different species of reptiles and amphibians., focusing on their ecological functions. Advanced deep learning techniques are applied to greatly improve the accuracy and efficiency of image-based species identification against these challenges due to the various appearances and subtle variations among these animals. The constructed classification model reached an accuracy of 84 % for all rounds., with Crocodile-Alligators and Turtle- Tortoises performing at their best and hitting precision scores of 0.95 and 0.96., respectively. Conversely., the poorest were Chameleons and Geckos., which obtained 0.76 and 0.66. All macro-averaged measures of precision., recall., and Fl-scores are very close to each other and come out to 0.79., 0.78., and 0.78., respectively., thus proving that the model performs quite well across different classes. Results indicate that CNN and transfer learning hold immense potential to increase the classification of reptiles and amphibians. This can help in monitoring and conserving biodiversity. A reliable tool for species identification will contribute towards greater understanding and protecting these vital components of our ecosystems., ultimately aiding the health and stability assessments of environmental conditions.