| Title: Investigating different spectrophotometry methods to detect weeds from agricultural products |
| Paper ID : 1402-NICAME1402 |
| Authors: |
|
Hossein Behfar *1, Milad Zamani2, Mahsa Javani2, Zahra Khadem2, Arezu Nobakht1 1Department of Biosystems Engineering, University of Tabriz, Tabriz, Iran 2biosystem engineering dep-Agriculture college-Tabriz university-Tabriz-Iran |
| Abstract: |
| The presence of weeds in the field with the crop creates many problems for agriculture. If it is possible to provide a method that is able to discriminant and distinguish the crop from the weed, a very basic challenge of farmers can be solved. Today, spectrophotometry methods are among the most important and widely used methods in this field, and their use has increased exponentially. Therefore, in this study, the aim is to investigate different spectrophotometry methods used for weed detection. In this study, the methods of detecting weed from the crop in different conditions and using different modeling method have been investigated. According to the results of this study, spectrophotometry methods in combination with pre-processing and different modeling methods can be successfully used to detect weed from the product. Among the spectrophotometry methods, the hyperspectral method and among the preprocessing methods, the normalization method, and convolutional neural networks method has often been more successful than other methods. |
| Keywords: |
| : Spectrophotometry, Weed, Detection |
| Status : Paper Accepted (Oral Presentation) |
