ICBI 2017
Permanent URI for this collectionhttp://repository.kln.ac.lk/handle/123456789/18303
Browse
Item Framework for Flower Gender Recognition Using Machine Learning.(8th International Conference on Business & Information ICBI – 2017, Faculty of Commerce and Management Studies, University of Kelaniya, Sri Lanka., 2017) De Silva, R.; Gunasinghe, H. N.This paper proposes a framework that can be used to identify the gender of imperfect flowers. One such application of gender identification of flowers is artificial pollination in large farmlands. The study reviews the literature on flower detection, flower recognition and its applications as well. Automatic gender identification of a flower is a branch of flower recognition that the researchers have not considered yet. The challenge in any automatic flower gender identification method is that the accuracy should be nearly 100 percent, as the maximum error rate of pollination attempts is twice that of identification. Our framework is based on building mathematical models of the structure of floral organs of imperfect flowers. It uses low-resolution images captured through cameras on aerial or mobile robots. Finally, it proposes to apply image processing and machine learning models together with image classification techniques to identify the gender of a given imperfect flower.Item Identification of Water Stress of Plants Using Image Processing.(8th International Conference on Business & Information ICBI – 2017, Faculty of Commerce and Management Studies, University of Kelaniya, Sri Lanka., 2017) De Silva, R.; Senanayake, P. A.Low Leaf Water Content (LWC) levels in plants lead to water stress, and the leaves become wilted. Most farmers use slight wilting as the visual indicator to water their plants. Though the researchers have studied LWC of plants extensively using remote sensing and complex methods, the agricultural industry cannot use them as they are complex and costly. This paper presents a simple method of recognizing water stress of leaves using leaf images taken by a smartphone. Our initial experiment on a large sample of mung bean leaves indicates that RGB values of images are related to water stress. The method would be beneficial to the agricultural industry as once further developed; it could be used to determine the watering time point of plants. The process can be automated by capturing the images by a camera mounted on a land or an aerial robot and processing them online.