발간년도 : [2020]
| 논문정보 |
|
| 논문명(한글) |
[Vol.15, No.2] Indoor Plants Image Classification Using Deep Learning and Web Application for Providing Information of Plants |
|
| 논문투고자 |
Hong-Jae Shin, So-In Lee, Hui-Won Jeoung, Jang-Woo Park |
|
| 논문내용 |
Plants have good effects such as air purification and landscaping, but they have special ingredients to protect themselves. Ingredients made to protect the plant itself can harm people or animals. There are also accidents that are mistaken for other plants with similar plant features. We have implemented a program that can identify plants and display information about each plant. Using deep learning to classify images, we created a web application that predicts plant names and displays information that matches the predicted plants. We created a data set of 61 indoor plants using Google Image Search. The positive effects and negative toxicity of indoor plants are summarized in the database. Deep learning is implemented using fast.ai, a Pytorch-based framework. Through data Augmentation, we increased the number of images to learn. Indoor plant image data were trained using ResNet50, a pretrained model using various images. The accuracy of the model was about 97.5%, which predicted most plants accurately. The web application was implemented using flask, a Python-based web framework. Using the implemented image classification deep learning model, the plant name is predicted and the information corresponding to the predicted plant name is displayed on the web page. The web application can be optimized for mobile devices and used conveniently. |
|
| 첨부논문 |
|
|
|
|
|
자료실 > 논문지 
