A Motion-Based Compression and Tracking System for Video Camera Trap-Based Insect Behaviour Studies

Abstract Field-captured video enables detailed study of animal locomotion, decision-making, and environmental interactions such as predator–prey dynamics and habitat use. While low-cost hardware makes data capture accessible, the storage, processing, and transmission demands of high-resolution video remain a major hurdle for field-deployed edge computing devices. Motion tracking in natural environments presents unique challenges that require tailored video compression strategies not well addressed in other domains. We present a novel end-to-end system comprising a motion analysis-based video compression algorithm specifically designed for camera traps, and a custom video processing methodology for automated analysis of compressed footage to extract behavioural data. We evaluate it through a case study on insect–pollinator motion tracking using three popular edge computing platforms. The compression algorithm operates alongside standard codecs, identifying and storing only image regions containing motion relevant to monitoring tasks, reducing data size by an average of 87% across diverse datasets. When combined with the H.265/HEVC codec, our approach achieved an additional 47.1% improvement in compression compared to stand-alone H.265. The accompanying video processing algorithm builds upon existing Polytrack software, incorporating new preprocessing and trajectory reconstruction techniques for automated processing of compressed footage with a 97.5% detection rate. Our experiments demonstrate that the system retains critical behavioural information, as verified through both automated and manual analyses. The method presented in this paper enhances the applicability of low-powered computer vision edge devices to remote, in situ animal motion monitoring, and improves the efficiency of playback during behavioural analyses.