| Popis: |
This research focuses on developing an architecture for fungus detection in fields using aerial imagery obtained from a quadcopter. Plant diseases, especially fungus, damages crops, reduce yield and lower the quality of produce. Manual inspection of plants in large fields by a farmer is a challenging and time-consuming task where sometimes it is not convenient to go deep in the fields. In this paper, fungus detection in fields using aerial imagery has been proposed, eliminating the need for manual inspection. The proposed method records the fields’ visual data by flying a quadcopter equipped with an RGB camera. After gathering the necessary images, the approach pre-processes the data and uses a Convolutional Neural Network (CNN) for training and classification. A dataset of more than 250, 000 images of gladiolus plants has been collected in this research work. The validity of the developed architecture and its usage in fields shows promising results with an accuracy of up to \(91\%\). |
