Hybrid ResNet-YOLOv5 Based Deep Learning Framework for Animal Monitoring in Ecosystem

Precise and effective animal monitoring is essential for sustaining environmental equilibrium and wildlife preser-vation. This research introduces an innovative Hybrid ResNet-YOLOv5 deep learning (DL) architecture for real-time moni-toring of animals in various environments. The system utilizes ResNet's residual learning for feature extraction and combines it with the YOLOv5 object detection model for reliable identification and tracking of animals in their natural environments. The system is engineered to accommodate many species and environmental circumstances, providing reliable performance in difficult situations such as occlusions and poor illumination. Experimental findings indicate that the Hybrid ResNet-YOLOv5 architecture attains an accuracy of 95.4%, surpassing standard ResNet(92.1%), YOLO (90.6%), and LSTM (88.9%) models. The suggested technique exhibits exceptional precision (94.5%), recall (96.2%), and Fl-score (95.3%), rendering it highly effective for wildlife monitoring and conservation efforts.