Detection of Wild Animal Activity Using Deep Learning Techniques: A Review
Author : Dr. T Suvarna Kumari, Sree Pratiksha Sanjivalla and Jayendra Medoju
Abstract :
Human-wildlife conflicts and the need for conservation have made efficient wildlife monitoring systems increasingly important. Traditional animal detection methods often rely on high-resolution, multi-frame imaging, which demands substantial computational power and is often expensive, particularly in remote and resource-constrained environments. Addressing these challenges, our project introduces a deep learning model designed to detect wild animals in images with both accuracy and efficiency. By combining the VGG-19 Convolutional Neural Network (CNN) for spatial feature extraction and a Bidirectional Gated Recurrent Unit (Bi-GRU) for temporal pattern recognition. Additionally, we employ YOLOv5 for precise localization, helping track animal positions within each frame. This combined model enhances detection capabilities, reduces computational demands, and lowers costs, making it ideal for deployment in remote areas with limited resources. Our solution is designed to contribute to real-time wildlife monitoring, aiding in conservation efforts and minimizing human-wildlife conflicts by providing a reliable, low-cost detection system that serves both environmental and community needs.
Keywords :
VGG-19 Convolutional Neural Network (CNN), Bidirectional Gated Recurrent Unit (Bi-GRU), YOLOv5.
