<![CDATA[Accurate wildlife identification in unconstrained environments remains a significant challenge due to intra-class similarity, occlusions, illumination variations, and complex backgrounds. This study of hybrid deep learning architecture that enhances classification performance by integrating complementary feature learning mechanisms within a unified framework. The model captures both deep hierarchical representations and multi-scale spatial features through a feature-level fusion strategy, resulting in a more discriminative and robust representation of wildlife images. The hybrid approach was trained and evaluated on the Animals with Attributes 2 (AwA2) dataset and further validated on the WildlifeReID-10k dataset to assess its generalization capability. Experimental results show that the model achieves a classification accuracy of 98%, along with high precision, recall, and F1-score, demonstrating strong reliability and effectiveness. Furthermore, the model maintains high performance on external data, confirming its robustness and ability to generalize across diverse real-world conditions. These findings highlight the effectiveness of hybrid deep learning approaches in addressing complex wildlife identification challenges and support their application in ecological monitoring and biodiversity conservation]]>
Copyrights © 2026
