Species recognition is an essential step in understanding animal behavior. Rather than manually identifying these species, we have developed software that allows us to classify the species in the submitted images semi-automatically. The output is, therefore, the assignment of a "label/labels" with the species name to the submitted photographs, by which further filtering can be done. The output is used within an existing workflow that uses manual tagging of photo trap images. The software will contribute to the improvement and refinement of tagging. 

December 2022

The current workflow for the manual selection of images from camera traps involves discarding images without the game present. To improve and refine the labelling, the software that performs the discarding semi-automatically has been developed. The input images are divided into three groups - an image containing an animal, an empty image, and an image that needs to be decided by a human.

September 2022

This software allows individual identification of various species in the wild. Besides, a submodule for the identification of Lynx lynx is provided. Given the sequences of images and appropriate location data, the system provides the Top-3 visually similar individuals, therefore, allowing a more reliable identification which consequentially enhances the efforts related to large carnivore protection.

December 2023