Recognition of Endemic Bird Species Using Deep Learning Models

Numerous bird species have become extinct because of anthropogenic activities and climate change. The destruction habitats at a rapid pace is significant threat to biodiversity worldwide. Thus, monitoring the distribution identifying elements that make up region are essential for designing conservation stratagems. However, from images complicated tedious task owing interclass similarities fine-grained features. To overcome this, in this study, we developed transfer learning-based method using Inception-ResNet-v2 detect classify endemic Taiwan distinguish them other object domains. Furthermore, validate reliability model, adopted technique involves swapping misclassified data between training validation datasets. swapped retrained until most suitable result obtained. Additionally, fivefold cross-validation was performed verify predictive performance model. proposed model tested 760 birds belonging 29 Taiwan; were captured various environments with different perspectives occlusions. Our achieved an accuracy 98.39% classification 100% detection among categories. Moreover, precision, recall, F1-score 98.49%, 97.50%, 97.90%, respectively, classifying Taiwan.