Detection of mite infested saffron plants using aerial imaging and machine learning classifier
Aim of study: To evaluate and develop a machine learning code that uses aerial images in visible and near infrared (NIR) spectra to detect mite-infested Saffron (Crocus sativus L.) plants through processing the spectral indices to classify healthy and diseased plants. This leads to the identificati...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Consejo Superior de Investigaciones Científicas (CSIC)
2025-01-01
|
| Series: | Spanish Journal of Agricultural Research |
| Subjects: | |
| Online Access: | https://sjar.revistas.csic.es/index.php/sjar/article/view/20452 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832576167848181760 |
|---|---|
| author | Hossein Sahabi Jalal Baradaran-Motie |
| author_facet | Hossein Sahabi Jalal Baradaran-Motie |
| author_sort | Hossein Sahabi |
| collection | DOAJ |
| description |
Aim of study: To evaluate and develop a machine learning code that uses aerial images in visible and near infrared (NIR) spectra to detect mite-infested Saffron (Crocus sativus L.) plants through processing the spectral indices to classify healthy and diseased plants. This leads to the identification of the concentration points of the bulb mites and the estimation of the percentage of infestation in the field.
Area of study: Khorasan-Razavi province, Torbat-Heydarieh, Iran.
Material and methods: Five fields were randomly selected and their red-green-blue (RGB), as a typical visible spectral image, and NIR images were taken in two consecutive years. Seven spectral vegetation indices for NIR images including NIR-band, Red-band, normalized difference vegetation index (NDVI), ratio vegetation index (RVI), difference vegetation index (DVI), difference red-nir ratio (DRN) and infrared percentage vegetation index (IPVI); and twelve indices for RGB images inlcuding red-band, green-band, blue-band, visible-band difference vegetation index (VDVI), visible atmospheric resistant index (VARI), triangular greenness index (TGI), normalized difference greenness index (NDGI), normalized green blue difference index (NGBDI), modified green red vegetation index (MGRVI), red green blue vegetation index (RGBVI), vegetative index (VEG) and excess of green index (EXG), were extracted and analysed. In order to detect affected plants, two support vector machine (SVM) classifiers with radial basis function (RBF) kernels were used separately for NIR and RGB images.
Main results: The average accuracy of the SVM classifier models were estimated to be 82.3% for NIR images and 91.4% for RGB images during the test phase. Also, the accuracy of the developed models when evaluated in the field with respect to the confusion matrix method was 75.6% and 80.3% for the classification models for NIR and RGB images, respectively.
Research highlights: RGB images were able to distinguish infested plants with better accuracy. Processing aerial images of lightweight drones could speed up the inspection of vast saffron fields.
|
| format | Article |
| id | doaj-art-ef4dd876a8c64d5f9e637accfe1a09a6 |
| institution | Kabale University |
| issn | 1695-971X 2171-9292 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Consejo Superior de Investigaciones Científicas (CSIC) |
| record_format | Article |
| series | Spanish Journal of Agricultural Research |
| spelling | doaj-art-ef4dd876a8c64d5f9e637accfe1a09a62025-01-31T10:29:47ZengConsejo Superior de Investigaciones Científicas (CSIC)Spanish Journal of Agricultural Research1695-971X2171-92922025-01-0122410.5424/sjar/2024224-20452Detection of mite infested saffron plants using aerial imaging and machine learning classifierHossein Sahabi0Jalal Baradaran-Motie1Department of Plant Production, University of Torbat Heydarieh, Torbat Heydarieh, 9516168595 Iran / Saffron Institute, University of Torbat Heydarieh, Torbat Heydarieh, 9516168595 IranDepartment of Biosystems Engineering, Ferdowsi University of Mashhad, Mashhad, 9177948974 Iran Aim of study: To evaluate and develop a machine learning code that uses aerial images in visible and near infrared (NIR) spectra to detect mite-infested Saffron (Crocus sativus L.) plants through processing the spectral indices to classify healthy and diseased plants. This leads to the identification of the concentration points of the bulb mites and the estimation of the percentage of infestation in the field. Area of study: Khorasan-Razavi province, Torbat-Heydarieh, Iran. Material and methods: Five fields were randomly selected and their red-green-blue (RGB), as a typical visible spectral image, and NIR images were taken in two consecutive years. Seven spectral vegetation indices for NIR images including NIR-band, Red-band, normalized difference vegetation index (NDVI), ratio vegetation index (RVI), difference vegetation index (DVI), difference red-nir ratio (DRN) and infrared percentage vegetation index (IPVI); and twelve indices for RGB images inlcuding red-band, green-band, blue-band, visible-band difference vegetation index (VDVI), visible atmospheric resistant index (VARI), triangular greenness index (TGI), normalized difference greenness index (NDGI), normalized green blue difference index (NGBDI), modified green red vegetation index (MGRVI), red green blue vegetation index (RGBVI), vegetative index (VEG) and excess of green index (EXG), were extracted and analysed. In order to detect affected plants, two support vector machine (SVM) classifiers with radial basis function (RBF) kernels were used separately for NIR and RGB images. Main results: The average accuracy of the SVM classifier models were estimated to be 82.3% for NIR images and 91.4% for RGB images during the test phase. Also, the accuracy of the developed models when evaluated in the field with respect to the confusion matrix method was 75.6% and 80.3% for the classification models for NIR and RGB images, respectively. Research highlights: RGB images were able to distinguish infested plants with better accuracy. Processing aerial images of lightweight drones could speed up the inspection of vast saffron fields. https://sjar.revistas.csic.es/index.php/sjar/article/view/20452Aerial imagingClassificationCrocus sativusImage processingSupport vector machine |
| spellingShingle | Hossein Sahabi Jalal Baradaran-Motie Detection of mite infested saffron plants using aerial imaging and machine learning classifier Spanish Journal of Agricultural Research Aerial imaging Classification Crocus sativus Image processing Support vector machine |
| title | Detection of mite infested saffron plants using aerial imaging and machine learning classifier |
| title_full | Detection of mite infested saffron plants using aerial imaging and machine learning classifier |
| title_fullStr | Detection of mite infested saffron plants using aerial imaging and machine learning classifier |
| title_full_unstemmed | Detection of mite infested saffron plants using aerial imaging and machine learning classifier |
| title_short | Detection of mite infested saffron plants using aerial imaging and machine learning classifier |
| title_sort | detection of mite infested saffron plants using aerial imaging and machine learning classifier |
| topic | Aerial imaging Classification Crocus sativus Image processing Support vector machine |
| url | https://sjar.revistas.csic.es/index.php/sjar/article/view/20452 |
| work_keys_str_mv | AT hosseinsahabi detectionofmiteinfestedsaffronplantsusingaerialimagingandmachinelearningclassifier AT jalalbaradaranmotie detectionofmiteinfestedsaffronplantsusingaerialimagingandmachinelearningclassifier |