Evaluation of Machine Learning Models for Stress Symptom Classification of Cucumber Seedlings Grown in a Controlled Environment
Stress by unfavorable environmental conditions, including temperature, light intensity, and photoperiod, significantly impact early-stage growth in crops, such as cucumber seedlings, often resulting in yield reduction and quality degradation. Advanced machine learning (ML) models combined with image...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-12-01
|
| Series: | Agronomy |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-4395/15/1/90 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832589386763468800 |
|---|---|
| author | Kyu-Ho Lee Samsuzzaman Md Nasim Reza Sumaiya Islam Shahriar Ahmed Yeon Jin Cho Dong Hee Noh Sun-Ok Chung |
| author_facet | Kyu-Ho Lee Samsuzzaman Md Nasim Reza Sumaiya Islam Shahriar Ahmed Yeon Jin Cho Dong Hee Noh Sun-Ok Chung |
| author_sort | Kyu-Ho Lee |
| collection | DOAJ |
| description | Stress by unfavorable environmental conditions, including temperature, light intensity, and photoperiod, significantly impact early-stage growth in crops, such as cucumber seedlings, often resulting in yield reduction and quality degradation. Advanced machine learning (ML) models combined with image-based analysis offer promising solutions for precise, non-invasive stress monitoring. This study aims to classify environmental stress symptom levels in cucumber seedlings using ML models by extracting critical color, texture, and morphological features from RGB images. In a controlled plant factory setup, two-week-old cucumber seedlings were subjected to varied environmental conditions across five chambers with differing temperatures (15, 20, 25, and 30 °C), light intensities (50, 250, and 450 µmol m<sup>−2</sup> s<sup>−1</sup>), and day-night cycles (8/16, 10/14, and 16/8 h). A cost-effective RGB camera, integrated with a microcontroller, captured images from the top of the seedlings over a two-week period, from which sequential forward floating selection (SFFS) and correlation matrices were used to streamline feature extraction. Four ML classifiers: Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Naïve Bayes (NB), and Random Forest (RF), were trained to detect stress symptoms based on selected features, highlighting that stress symptoms were detectable after day 4. KNN achieved the highest accuracy at 0.94 (94%), followed closely by SVM and RF, both at 93%, while NB reached 88%. Findings suggested that color and texture features were critical indicators of stress, and that the KNN model, with optimized hyperparameters, provided a reliable classification for stress symptom monitoring for seedlings under controlled environments. This study highlights the potential of ML-driven stress symptom detection models for controlled seedling production, enabling real-time decision-making to optimize crop health and productivity. |
| format | Article |
| id | doaj-art-321cf2fc8c1a4ae39bb3c701f144a206 |
| institution | Kabale University |
| issn | 2073-4395 |
| language | English |
| publishDate | 2024-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Agronomy |
| spelling | doaj-art-321cf2fc8c1a4ae39bb3c701f144a2062025-01-24T13:16:41ZengMDPI AGAgronomy2073-43952024-12-011519010.3390/agronomy15010090Evaluation of Machine Learning Models for Stress Symptom Classification of Cucumber Seedlings Grown in a Controlled EnvironmentKyu-Ho Lee0Samsuzzaman1Md Nasim Reza2Sumaiya Islam3Shahriar Ahmed4Yeon Jin Cho5Dong Hee Noh6Sun-Ok Chung7Department of Agricultural Machinery Engineering, Graduate School, Chungnam National University, Daejeon 34134, Republic of KoreaDepartment of Agricultural Machinery Engineering, Graduate School, Chungnam National University, Daejeon 34134, Republic of KoreaDepartment of Agricultural Machinery Engineering, Graduate School, Chungnam National University, Daejeon 34134, Republic of KoreaDepartment of Smart Agricultural Systems, Graduate School, Chungnam National University, Daejeon 34134, Republic of KoreaDepartment of Agricultural Machinery Engineering, Graduate School, Chungnam National University, Daejeon 34134, Republic of KoreaJeonnam Agricultural Research and Extension Services, Naju 58213, Republic of KoreaJeonbuk Regional Branch, Korea Electronics Technology Institute (KETI), Jeonju 54853, Republic of KoreaDepartment of Agricultural Machinery Engineering, Graduate School, Chungnam National University, Daejeon 34134, Republic of KoreaStress by unfavorable environmental conditions, including temperature, light intensity, and photoperiod, significantly impact early-stage growth in crops, such as cucumber seedlings, often resulting in yield reduction and quality degradation. Advanced machine learning (ML) models combined with image-based analysis offer promising solutions for precise, non-invasive stress monitoring. This study aims to classify environmental stress symptom levels in cucumber seedlings using ML models by extracting critical color, texture, and morphological features from RGB images. In a controlled plant factory setup, two-week-old cucumber seedlings were subjected to varied environmental conditions across five chambers with differing temperatures (15, 20, 25, and 30 °C), light intensities (50, 250, and 450 µmol m<sup>−2</sup> s<sup>−1</sup>), and day-night cycles (8/16, 10/14, and 16/8 h). A cost-effective RGB camera, integrated with a microcontroller, captured images from the top of the seedlings over a two-week period, from which sequential forward floating selection (SFFS) and correlation matrices were used to streamline feature extraction. Four ML classifiers: Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Naïve Bayes (NB), and Random Forest (RF), were trained to detect stress symptoms based on selected features, highlighting that stress symptoms were detectable after day 4. KNN achieved the highest accuracy at 0.94 (94%), followed closely by SVM and RF, both at 93%, while NB reached 88%. Findings suggested that color and texture features were critical indicators of stress, and that the KNN model, with optimized hyperparameters, provided a reliable classification for stress symptom monitoring for seedlings under controlled environments. This study highlights the potential of ML-driven stress symptom detection models for controlled seedling production, enabling real-time decision-making to optimize crop health and productivity.https://www.mdpi.com/2073-4395/15/1/90controlled environment agriculturemachine learningcucumber seedlingimage analysisstress symptom classification |
| spellingShingle | Kyu-Ho Lee Samsuzzaman Md Nasim Reza Sumaiya Islam Shahriar Ahmed Yeon Jin Cho Dong Hee Noh Sun-Ok Chung Evaluation of Machine Learning Models for Stress Symptom Classification of Cucumber Seedlings Grown in a Controlled Environment Agronomy controlled environment agriculture machine learning cucumber seedling image analysis stress symptom classification |
| title | Evaluation of Machine Learning Models for Stress Symptom Classification of Cucumber Seedlings Grown in a Controlled Environment |
| title_full | Evaluation of Machine Learning Models for Stress Symptom Classification of Cucumber Seedlings Grown in a Controlled Environment |
| title_fullStr | Evaluation of Machine Learning Models for Stress Symptom Classification of Cucumber Seedlings Grown in a Controlled Environment |
| title_full_unstemmed | Evaluation of Machine Learning Models for Stress Symptom Classification of Cucumber Seedlings Grown in a Controlled Environment |
| title_short | Evaluation of Machine Learning Models for Stress Symptom Classification of Cucumber Seedlings Grown in a Controlled Environment |
| title_sort | evaluation of machine learning models for stress symptom classification of cucumber seedlings grown in a controlled environment |
| topic | controlled environment agriculture machine learning cucumber seedling image analysis stress symptom classification |
| url | https://www.mdpi.com/2073-4395/15/1/90 |
| work_keys_str_mv | AT kyuholee evaluationofmachinelearningmodelsforstresssymptomclassificationofcucumberseedlingsgrowninacontrolledenvironment AT samsuzzaman evaluationofmachinelearningmodelsforstresssymptomclassificationofcucumberseedlingsgrowninacontrolledenvironment AT mdnasimreza evaluationofmachinelearningmodelsforstresssymptomclassificationofcucumberseedlingsgrowninacontrolledenvironment AT sumaiyaislam evaluationofmachinelearningmodelsforstresssymptomclassificationofcucumberseedlingsgrowninacontrolledenvironment AT shahriarahmed evaluationofmachinelearningmodelsforstresssymptomclassificationofcucumberseedlingsgrowninacontrolledenvironment AT yeonjincho evaluationofmachinelearningmodelsforstresssymptomclassificationofcucumberseedlingsgrowninacontrolledenvironment AT dongheenoh evaluationofmachinelearningmodelsforstresssymptomclassificationofcucumberseedlingsgrowninacontrolledenvironment AT sunokchung evaluationofmachinelearningmodelsforstresssymptomclassificationofcucumberseedlingsgrowninacontrolledenvironment |