How do you find the green sheep? A critical review of the use of remotely sensed imagery to detect and count animals

• Published in: Methods in Ecology and Evolution
• Main Authors: Hollings, Tracey, Burgman, Mark, van Andel, Mary, Gilbert, Marius, Robinson, Timothy, Robinson, Andrew
• Other Authors: McPherson, Jana
• Format: Article in Journal/Newspaper
• Language: English
• Published: Wiley 2018
• Subjects: Sea ice
• Online Access: http://dx.doi.org/10.1111/2041-210x.12973; https://api.wiley.com/onlinelibrary/tdm/v1/articles/10.1111%2F2041-210X.12973; https://onlinelibrary.wiley.com/doi/pdf/10.1111/2041-210X.12973; https://onlinelibrary.wiley.com/doi/full-xml/10.1111/2041-210X.12973; https://besjournals.onlinelibrary.wiley.com/doi/pdf/10.1111/2041-210X.12973

Abstract Animal abundance data are essential for endangered species conservation, tracking invasive species spread, biosecurity, agricultural applications and wildlife monitoring; however, obtaining abundance data are a perennial challenge. Recent improvements in the resolution of remotely sensed imagery, and image‐processing tools and software have facilitated improvement of methods for the detection of individual, generally large‐bodied animals. The potential to monitor and survey populations from remotely sensed imagery is an exciting new development in animal ecology. We review the methods used to analyse remotely sensed imagery for their potential to estimate the abundance of wild and domestic animal populations by directly detecting, identifying and counting individuals. Despite many illustrative studies using a variety of methods for detecting animals from remotely sensed imagery, it remains problematic in many situations. Studies that demonstrated reasonably high accuracy using automated and semi‐automated techniques have been undertaken on small spatial scales relative to the geographical range of the species of interest and/or in homogenous environments such as sea ice. The major limitations are the relatively low accuracy of automated detection techniques across large spatial extents, false detections and the cost of high‐resolution data. Future developments in the analysis of remotely sensed data for population surveys will improve detection capabilities, including the advancement of algorithms, the crossover of software and technology from other disciplines, and improved availability, accessibility, cost and resolution of data.