A Hierarchical Model for Classifying Sea Turtle Behavior from Sensor Data

Bio-logging technology facilitates remote monitoring of animal movement patterns. This study introduces a hierarchical classification framework for identifying behavioral patterns from the Inertial Measurement Unit data, including accelerometer and gyroscope measurements, collected from green sea turtles. The classifier utilizes a two-tier architecture that first categorizes broad behavioral groups, then performs detailed discrimination of specific behavioral classes. Experimental results show 92% accuracy in classifying 49 distinct behavioral categories.