Tree Counting with Drones
Problem being addressed
Urban forestry plays a number of roles including shade, carbon storage and food. However, given the complexity of urban spaces, it can be hard to keep track of the tree count. How can a drone and AI do the job?
Multirotor UAVs equipped with high resolution cameras can be used to take precise images of the citiscape. In this research an imaging filter algorithm is created to sample the images from a UAV and produce a tree count as output. The basic process is to use a series of filters on the raw image data starting with an entropy filter, and then following up with a colour based clustering technique, and finally a circle fitting algorithm which enables the tree count.
Advantages of this solution
Tree counting has an important role to play in tracking urban carbon sequestration projects. This research provides a cheap and scalable way to do enable the process.
Solution originally applied in these industries
Possible New Application of the Work
There are obvious direct applications of this research in forestry and other branches of agriculture, but perhaps the images could go into finer detail and understand plant health by, for example, counting lichen on the trees themselves? The presence or absence of lichen is usually a good indicator of air pollution levels.
The basic premise of this research is counting from image data. One application to the retail industry would be to count foot traffic in and out of stores and malls. In this way, shops could optimize their staff allocation to peak times.
Transport and Logistics Industry
Traffic monitoring systems could be designed in a similar way to this tree counting system by using fixed cameras and an algorithm to count cars. This could then be linked in real-time to traffic signals in order to regulate the flow of traffic.
Source DOI: #############
Source URL: #############