AlexNet-Based Feature Extraction for Cassava Classification: A Machine Learning Approach

Extractor Feature (linguistics)
DOI: 10.21123/bsj.2023.9120 Publication Date: 2023-12-05T09:21:51Z
ABSTRACT
Cassava, a significant crop in Africa, Asia, and South America, is staple food for millions. However, classifying cassava species using conventional color, texture, shape features inefficient, as leaves exhibit similarities across different types, including toxic non-toxic varieties. This research aims to overcome the limitations of traditional classification methods by employing deep learning techniques with pre-trained AlexNet feature extractor accurately classify four types cassava: Gajah, Manggu, Kapok, Beracun. The dataset was collected from local farms Lamongan Indonesia. To collect images agricultural experts, consists 1,400 images, each type has 350 images. Three fully connected (FC) layers were utilized extraction, namely fc6, fc7, fc8. classifiers employed support vector machine (SVM), k-nearest neighbors (KNN), Naive Bayes. study demonstrated that most effective extraction layer achieving an accuracy 90.7% SVM. SVM outperformed KNN Bayes, exhibiting 90.7%, sensitivity 83.5%, specificity 93.7%, F1-score 83.5%. successfully addressed challenges leveraging methods, specifically fc6 AlexNet. proposed approach holds promise enhancing plant techniques, benefiting researchers, farmers, environmentalists identification, ecosystem monitoring, management.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (0)
CITATIONS (0)