Towards detection of animals in aerial photography

Speaker: Chrissy Fell (University of St.Andrews)

Abstract

Surveys of animals in large areas of land can be carried out using aerial photography. The current bottleneck with this process is that the large number of photographs generated then have to be viewed by people to find the animals. This is often so time consuming and tedious that it reduces the use of this method of surveying. Modern computer vision techniques hold out the hope that this could now be automated. In this talk, I will discuss an aerial animal detection dataset and application of techniques such as segmentation, HOG and CNNs to detecting animals.