A novel method to reduce time investment when processing videos from camera trap studies

Area of Science:

• Ecology
• Conservation Biology
• Wildlife Behavior

Background:

• Camera traps are essential tools for non-invasive ecological and behavioral research, generating vast amounts of data.
• Current data processing protocols lag behind camera trap technological advancements, leading to significant time investment.
• Manual removal of non-target recordings (empty or containing other species) is a major bottleneck in camera trap data analysis.

Purpose of the Study:

• To develop and evaluate a method for automatically filtering non-target recordings from camera trap video data.
• To reduce the workload associated with processing large volumes of camera trap footage, specifically for Eurasian beaver (Castor fiber) research.
• To improve the efficiency of wildlife research workflows by automating the initial stages of data processing.

Main Methods:

• Proposed a method to discriminate between target (Eurasian beaver) and non-target recordings based on frame-to-frame pixel variation.
• Assumed that recordings with the target species would exhibit greater movement than non-target recordings.
• Tested and compared two different filter methods to identify and discard non-target recordings.

Main Results:

• Demonstrated that variation in pixel values can be used to partially discriminate between target and non-target recordings.
• Found that environmental conditions and specific filter methods impact the efficiency of non-target recording identification.
• Achieved identification and discarding of 53% to 76% of non-target recordings with a 5% to 20% loss of target species recordings.

Conclusions:

• Implementing an automated filtering step in camera trap protocols can lead to substantial time savings for researchers.
• The proposed method offers a valuable tool for researchers globally, applicable to both video and photographic data across various wildlife research contexts.
• This approach addresses the growing challenge of managing large datasets in camera trap studies, enhancing research efficiency and scope.