Development of Android Application for Plant Symbiosis Recognition Integrating Convolutional Neural Network

Since every plant has a different set of needs and may have positive or negative effects on one another, selecting the right plants is important. In this study, an android application which uses Convolutional Neural Network for image classification was developed. Images of various vegetable leaves were manually taken with a camera and downloaded from the Kaggle website for the dataset. In comparison to the manually taken photographs, which numbered about 7600, the images selected in Kaggle had about 3600 samples. There are 14 different types of vegetables, and there are a total of 11,200 photos. Every image is scaled to 224 × 224 pixels. The best model for plant detection was calibrated, optimized, and used to identify whether plant would be compatible with or complement other plants as part of the convolutional neural network integration process. The Convolutional Neural Network classification performance in identifying a vegetable using the confusion matrix resulted in scores of 94.82% accuracy. Hence, this application is reliable in plant detection and the application can be used to identify plant symbiosis. It was concluded that the application was evaluated as strongly satisfactory according to ISO 25010 software quality standards.