Peer-reviewed
Vision-based pest detection based on SVM classification method
• Detection of thrips in strawberry greenhouse. • Detection based on SVM classification method. • Pest detection using visible image processing. • Real-time detection. • Detection under the natural light.
Automatic pest detection is a useful method for greenhouse monitoring against pest attacks. One of the more harmful pests that threaten strawberry greenhouses is thrips (Thysanoptera). Therefore, the main objective of this study is to detect of thrips on the crop canopy images using SVM classification method. A new image processing technique was utilized to detect parasites that may be found on strawberry plants. SVM method with difference kernel function was used for classification of parasites and detection of thrips. The ratio of major diameter to minor diameter as region index as well as Hue, Saturation and Intensify as color indexes were utilized to design the SVM structure. Also, mean square error (MSE), root of mean square error (RMSE), mean absolute error (MAE) and mean percent error (MPE) were used for evaluation of the classification. Results show that using SVM method with region index and intensify as color index make the best classification with mean percent error of less than 2.25%
Automatic pest detection is a useful method for greenhouse monitoring against pest attacks. One of the more harmful pests that threaten strawberry greenhouses is thrips (Thysanoptera). Therefore, the main objective of this study is to detect of thrips on the crop canopy images using SVM classification method. A new image processing technique was utilized to detect parasites that may be found on strawberry plants. SVM method with difference kernel function was used for classification of parasites and detection of thrips. The ratio of major diameter to minor diameter as region index as well as Hue, Saturation and Intensify as color indexes were utilized to design the SVM structure. Also, mean square error (MSE), root of mean square error (RMSE), mean absolute error (MAE) and mean percent error (MPE) were used for evaluation of the classification. Results show that using SVM method with region index and intensify as color index make the best classification with mean percent error of less than 2.25%
Article, 2017