In the Netherlands it is mandatory that the 1st of November of every year all ditches are clean. This is inspected by “de waterschappen”. Can this laborious task can be automated using an Unmanned Aerial Vehicle with a computer on board, which can inspect the ditches and report whether they have to be cleaned or not? Therefore, the project will consist of tracking a ditch and inspecting it using image processing.
Most of the farmers inspect the state of their agricultural fields manually, which is not only an intensive and time consuming but also subjective task. This can be improved by taking pictures of the fields from the air. An Smart Civil Drone (or Unmanned Aerial Vehicle) with a computer on board can be used to take these pictures and automatically inspect the field using a hyperspectral or a multispectral camera (a camera which records more than the conventional three –red, green and blue- bands of the spectrum, in order to provide more information).
The goal is to detect potato plant anomalies and diseases from the images taken by an unmanned aerial vehicle. This consist of stitching, image analysis and data science.
Students: 1 or 2
The NHL Centre of Expertise in Computer Vision works on automating visual inspections using Smart Civil Drones (or Smart Unmanned Aerial Vehicles). The main goal is to research and develop a UAV which can automatically perform navigation tasks using computer vision and sensor fusion. For example: take off, approach and aligning with an object, hover at one position, return to the initial point, landing, line following and moving object following.
This enables applied research projects in a multitude of applications like wind turbine blade inspection and precision agriculture. Interest in learning C++ is required.
Students: 1 or 2 (Computer Science)
When fighting a fire it is very important to locate the fire spots and be able to make a risk analysis. A camera in the air can offer a good overview of the situation, but the images that provides have to be interpreted. The main goal is to automate the acquisition of images and interpret them using computer vision.
The information extracted can be mapped to get an overview of the different fire and smoke cores. This can be done using an Unmanned Aerial Vehicle (UAV) with a camera and a computer on board which automatically inspects the area and maps lively where these cores are.
Normal cameras take either grayscale or color images. A grayscale image is composed by pixels with different intensities, that go from black to white, while a color image has more information, for example the amount of red, green and blue of every pixel. Using computer vision, we can tell a lot about the world from grayscale or color images. Fire though, has another feature which can provide with a lot of information: the temperature. In order to get an image with temperature information in it, an innovative light weighted thermographic camera has been acquired. This camera is able to record the Long Wave Infrared (LWIR) region of the spectrum.
This project will consist of helping to interface this LWIR camera so it can be used with our computer vision software. This will be done mainly in C++, so it is required interest on learning this programming language.
Students: 1 or 2 (Computer Science)
For the maintenance of windturbines, it is essential to detect the defects of the blades. Even small defects on these blades reduce the efficiency of the whole turbine. Currently, the blades are inspected by trained people who have to climb up in the wind turbine and inspect them manually. Smart Civil Drones (or Smart Unmanned Aerial Vehicles) can be used to inspect the wind turbine blades.
The goal is to automatically scan the windmill blades and perform image analysis of the defects on the blades using image analysis and data science.