Improved Automated Recognition of Leopards from Photographs

Abstract

Two improvements on the automated image recognition of individual leopards (Panthera pardus) are presented. Firstly, a new method of automated segmentation on images, with a guaranteed side view of the animal, allows for a region of interest to be calculated automatically, reducing the manual effort of the image processing substantially. Secondly, improvements on the automated recognition of citizen science images with extreme viewpoints are detailed.