Development of Invasive Plant Recognition System Based on Deep Learning
Zhuolei Yang *
School of Mathematics and Statistics, The School of Software, The School of Computer and Information Engineering, Miami College, Henan University, Kaifeng City 475000, China.
Zheming Fan
North Section of Jinming Avenue, Kaifeng City, Henan Province, China.
Chenyu Niu
North Section of Jinming Avenue, Kaifeng City, Henan Province, China.
Peixin Li
North Section of Jinming Avenue, Kaifeng City, Henan Province, China.
Hongjie Zhong
North Section of Jinming Avenue, Kaifeng City, Henan Province, China.
*Author to whom correspondence should be addressed.
Abstract
A major ecological issue that has seriously harmed both human society and the environment is the invasion of alien plants. To stop the invasion of alien plants, it is crucial to create an effective and precise monitoring and early warning system. In this situation, deep learning and computer vision have significant potential to enhance plant monitoring on a wide scale. This study suggests a deep learning-based approach for identifying invasive plants. The user interface is developed as a mobile application (APP). The identification result can be acquired in 1 to 2 seconds after downloading the plant image from the APP, uploading it to the server, and using the convolutional neural network (CNN). The system had an average accuracy of 90.39% on the test set thanks to data augmentation and enhanced networks. The deep learning-based invasive plant identification system created in this study has demonstrated through experiments that it may effectively support botanical research and ecological environment monitoring.
Keywords: Alien plant, deep learning, mobile client, CNN, image recognition