Advancing primate surveillance with image recognition techniques from unmanned aerial vehicles

Area of Science:

• Wildlife monitoring
• Conservation technology
• Primate research

Background:

• Unmanned aerial vehicles (UAVs) offer practical and precise dynamic censuses for wildlife monitoring.
• Automating image interpretation is crucial for enhancing the effectiveness of UAV-based animal monitoring.
• Conservation strategies benefit from accurate population data derived from advanced monitoring techniques.

Purpose of the Study:

• To demonstrate the effective use of thermal cameras on UAVs for monitoring Sichuan snub-nosed monkey (Rhinopithecus roxellana) populations.
• To develop and validate an automated image interpretation method for primate population surveys.
• To establish a formula for adjusting primate population estimates based on UAV survey detection rates.

Main Methods:

• Utilized a local contrast method for a small infrared target detection algorithm to count the total population.
• Determined the average optimal grayscale threshold using an experimental group.
• Validated the automated detection and counting of target animals using a validation group with similar datasets.

Main Results:

• Achieved a precision rate ranging from 85.14% to 97.60% in automated detection.
• Identified a negative correlation between the minimum average distance between thermal spots and the count of detected individuals.
• Proposed a formula to adjust primate population estimates based on UAV survey detection rates.

Conclusions:

• UAV-based thermal imagery combined with automated detection algorithms offers a practical approach for primate monitoring.
• Environmental factors and data preprocessing are important considerations for accurate population estimates.
• This study advances UAV technology applications in wildlife monitoring, supporting conservation management and research.