Home
A system for taking using aerial footage of various types and performing both image stitching and satellite map overlays. This is to help in the effort to conserve the state of national parks in South Africa and allows a user to manage footage and maps associated with parks and provide secure access to imagery assets. Some advanced features include the generation of 3D point clouds from the 2D images, splitting videos into frames for stitching, georeferencing of images to be overlayed onto existing maps and object recognition to highlight artefacts of interest.
Nature conservation and anti-poaching initiatives are major focus areas in our society. From movement-sensing cameras to tracking collars on animals, technology plays a major role in the conservation of endangered animals. While the small satellite industry is doing extremely well in terms of earth imaging in higher and higher spatial and temporal resolution, they can not be used to resolve anything less than 0.5m per pixel currently. High-resolution aerial footage largely solves this issue but doesn’t have the same field of view as a satellite. In order to solve this problem, we plan on bridging this gap by taking multiple aerial images and using their identifiable features to overlay and join images into a detailed bigger picture. This will be optimised by either supervised or unsupervised image recognition machine learning techniques. This will allow for various analyses to be performed, such as snare and trap detection, localised water scarcity, changes in animal paths and tracking changes in vegetation health among others.
Drones allow for a more up-to-date set of images of a landscape, this allows much higher temporal resolution and can help mitigate many issues currently faced in conservation. A key mission in the project will be that of tracking changes in an area over time, so a large part of the project will be focused on optimising that core functionality.
