A HYBRID DEEP LEARNING FRAMEWORK FOR WILD ANIMAL DETECTION AND AUTOMATED ALERT MESSAGING

Abstract

Forestry workers and rural residents are becoming more and more concerned about the problem of animal assaults. Both drones and surveillance cameras are often used to monitor the movements of wild animals. To identify the sort of animal, track its movement, and offer its position, an effective model is necessary. The safety of both people and foresters may then be guaranteed by sending alert signals. Although methods based on computer vision and machine learning are widely employed for animal identification, their cost and complexity make it challenging to get adequate results. In order to identify animals and provide warnings depending on their behaviour, this research proposes a Hybrid Visual Geometry Group (VGG)−19+ Bidirectional Long Short-Term Memory (Bi-LSTM) network. To enable prompt action, these notifications are sent by Short Message Service (SMS) to the nearby forest office. With an average classification accuracy of 98%, a mean Average Precision (mAP) of 77.2%, and a Frame Per Second (FPS) of 170, the suggested model demonstrates significant gains in model performance. Using 40,000 photos from three distinct benchmark datasets with 25 classes, the model was evaluated both subjectively and statistically. Its mean accuracy and precision were higher than 98%. This approach is a dependable way to save human life and provide precise information based on animals.