Sensor Fusion Based Approach for Detection of Illegal Tree Felling

Illegal tree-felling is a significant threat to forest ecosystems and biodiversity. This paper presents a real-time detection system that integrates audio classification, which involves extracting audio features for supervised classification algorithm and integrating this with multisensor data to identify tree cutting activities. Each tree is equipped with an ESP32 module to monitor tilt and wire breakage, while a central Raspberry Pi processes sound signals using a K-Nearest Neighbors (KNN) algorithm to classify saw-cutting sounds from other possible environmental sounds. Sensor data are transmitted via the Message Queuing Telemetry Transport (MQTT) protocol, thereby facilitating centralized decision-making. An alert is triggered only when at least two of the three parameters (sound, tilt, and wire break) indicate cutting.