Shugo Nakamura

Industry Track: An Image Classification and Browsing System for Farm Inspection

When a farmer shares information about farm situation or events with other farmers, he often uses pictures (images) taken during farm inspection for visual communication. However, it may take much time for the farmers to find the desired images from a large amount of images accumulated every day. Furthermore, the contents of the images are diverse (e.g. crop images, soil images, and field images), and the content of the desired images are dependent on the usage scene. Therefore, we develop an image classification and browsing system to suitable for the usage scene. We adopt a typical SVM image classification method using color histogram or layout of brightness as image features. The effectiveness of our system is verified by feasibility study in checking the growth situation of crops. The contribution of this work is the first attempt to apply the image classification and browsing technique to the farm inspection support in agricultural fields.