Principal Investigator: Assistant Professor Henrik Skov Midtiby

Faculty of Engineering – Mærsk Mc-Kinney Møller Institute – SDU Dronecenter – SDU

We work on recognizing objects of interest in images from natural environments. A case is to detect the presence of giant hogweed plants in an image. We use convolutional neural networks (CNNs) to detect these plants. The use of CNN requires access to suitable computing resources, especially graphics processing units GPU. We now have a CNN that reliably can detect giant hogweed in images, the next step is then to limit the number of false positive detections by the CNN.