A motion-based compression algorithm for resource-constrained video camera traps

TL;DR

A new motion analysis-based video compression algorithm specifically designed for camera traps that 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.

Abstract

Field-captured video facilitates detailed studies of spatio-temporal aspects of animal locomotion, decision-making and environmental interactions including predator-prey relationships and habitat utilisation. But even though data capture is cheap with mass-produced hardware, storage, processing and transmission overheads provide a hurdle to acquisition of high resolution video from field-situated edge computing devices. Efficient compression algorithms are therefore essential if monitoring is to be conducted on single-board computers in situations where such hurdles must be overcome. Animal motion tracking in the field has unique characteristics that necessitate the use of novel video compression techniques, which may be underexplored or unsuitable in other contexts. In this article, we therefore introduce a new motion analysis-based video compression algorithm specifically designed for camera traps. We implemented and tested this algorithm using a case study of insect-pollinator motion tracking on three popular edge computing platforms. The algorithm identifies and stores only image regions depicting motion relevant to pollination monitoring, reducing overall data size by an average of 87% across diverse test datasets. Our experiments demonstrate the algorithm's capability to preserve critical information for insect behaviour analysis through both manual observation and automatic analysis of the compressed footage. 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. Our new software, EcoMotionZip, is available Open Access.

Figures (6)

• Figure 1: Overview of the algorithm architecture. The proposed algorithm has three main components: (1) Reader, (2) Motion Analysis, and (3) Writer, implemented with three separate threads. Footage from van2022continuous illustrates the algorithm.
• Figure 2: Frame pre-processing for compressed videos. This step processes video frames captured using a compression algorithm to make them compatible with Polytrack analysis ratnayake2023spatial. "For Deep Learning-based Detection" indicates the RGB image input to Polytrack to be used by the deep learning-based detection model and "For FGBG Detection" shows the image input to Polytrack to be used by the foreground-background segmentation model.
• Figure 3: Application environments in the test dataset. (a) naqvi2022camera, (b) Ratnayake2020_dataset, (c) Ratnayake2022_dataset, (d) van2022continuous, (e) Nest Monitoring, and (f) droissart2021pict
• Figure 4: Percentage of pixels changed per frame in test datasets. The percentage of pixels changed per frame represents the proportion of pixels with altered intensity values between two consecutive frames in the video dataset. The red diamonds indicate the mean values, and the orange line represents the median for each dataset.
• Figure 5: Insect trajectories and flower positions were extracted using automated video observations from (a) raw videos and (b) compressed videos. Video processing was conducted using Polytrack ratnayake2023spatial. In the legend, for each type of insect, "T" indicates the number of tracks recorded, "F" denotes the number of flower visits made Flower locations are highlighted with yellow circles.