Creating a behavioural classification module for acceleration data: Using a captive surrogate for difficult to observe species

• DOI: 10.1242/jeb.089805
• Publication Date: 2013-09-13T01:11:38Z
• Authors: Hamish Campbell; Lianli Gao; Owen Bidder; Jane Hunter; Craig Franklin
• Keywords: Binary classification

Summary Distinguishing specific behavioural modes from data collected by animal-borne tri-axial accelerometers can be a time consuming and subjective process. Data synthesis further inhibited when the acceleration cannot paired with corresponding mode through direct observation. Here we explored use of tame surrogate (domestic dog) to build classification module, then used that module accurately identify quantify within other individuals/species. Tri-axial were recorded domestic dog whilst it was commanded walk, run, sit, stand, lie-down. Through video synchronisation, each sample annotated its associated mode; feature vectors extracted, application support vector machines (SVM). This same in range species (alligator, badger, cheetah, dingo, echidna, kangaroo, wombat). Evaluation performance, using binary system, showed there high capacity (> 90 %) for behaviour recognition between individuals species. Furthermore, positive correlation existed SVM extent which individual had spinal length-to-height above ground ratio (SL:SH) similar surrogate. The study describes how highlights value studying cryptic, rare or endangered