Flower Based Plant Classification System

Plants could be classified using plant flower information. Lot of existing flower-based plant classification systems are available. Some classification system uses DSP techniques to classify the plants. Some classification systems use machine learning techniques to classify the plants. Some classification system uses hybrid techniques to classify the plants. Some classification system uses other methods to classify the plants. Although we have lot of flower-based plant classification systems, the accuracy of the existing flower-based plant classification systems is low. We use machine learning techniques to classify the plants. Our system uses length, width, and mean RGB values of the flowers as the features of the machine learning models. We use K Nearest Neighbour, Support vector machine, Random Forest, and CNN algorithms to classify the plants. We collected the data from five different plants in the Jaffna district, in Sri Lanka. The plants are Nithiyakalyani, Nanthiyavattai, Nochi, Thiruvaththi, and Hibiscus. Each class has 50 samples. The length and width are measured using measuring tape. The RGB mean value is measured using digital camera. The local patterns of the plant species could be found in the first layer of the convolutional neural network models.