Empirical Technique of Forest Conservation using Machine Learning and Cloud Computing
Veeranna Kotagi, Shob Raj Gowda K, D Vaibhav, Swaroop Kumar, Ajay Swamy · 2025
This paper proposes a machine learning and cloud computing-based system for forest conservation, focusing on sound classification to detect activities like illegal logging, wildlife movement, and human interference. The approach uses Mel-frequency cepstral coefficients (MFCCs), chroma features, and spectrograms for audio feature extraction. A Convolutional Neural Network (CNN) was trained on a diverse dataset of natural and anthropogenic sounds, achieving a macro-average precision of 0.76, recall of 0.69, and F1 score of 0.69, with 69% accuracy on 400 test samples. The model showed strong performance in recognizing critical sounds such as chainsaw activity, footsteps, and dog barking. Cloud services supported model training and data management, providing scalability and efficient resource use. The results demonstrate that integrating machine learning and cloud computing offers a practical and effective solution for forest conservation, enabling real-time sound-based threat detection and promoting proactive environmental protection efforts