Hybrid Color Feature Image Categorization using Machine Learning
| Autor: | M. Seshashayee, Shameem Fatima |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Contextual image classification
business.industry Computer science 020208 electrical & electronic engineering Feature extraction Sorting 04 agricultural and veterinary sciences 02 engineering and technology Machine learning computer.software_genre 040401 food science Support vector machine 0404 agricultural biotechnology Categorization Feature (computer vision) Kernel (statistics) Classifier (linguistics) 0202 electrical engineering electronic engineering information engineering Artificial intelligence business computer |
| Zdroj: | 2020 3rd International Conference on Intelligent Sustainable Systems (ICISS). |
| DOI: | 10.1109/iciss49785.2020.9316111 |
| Popis: | Agricultural products, particularly fruits, forms an important part of trading. The agriculture industry has gained attention in recent years for the purpose of categorizing the fruit crop based on their visual quality. Sorting and grading plays an important role among post harvesting operations. Traditional sorting was human-dependent, which is time consuming, less efficient due to labor shortage during peak seasons therefore automation of the fruits grading system is needed which in turn depend on external features. Several machine learning techniques for two class problem are applied for fruit image classification. In this paper Machine Learning is applied to multiclass fruit categorization using color feature. The improved model based on hybrid color feature is developed and presented to benefit agriculture industry to make the process of sorting easy for local vendors with color features. The training of model is performed using support vector machine (SVM) classifier on fruit dataset which includes 6 fruit categories. The result shows that the use of hybrid color feature with linear kernel technique with OneVsOne coding design enhanced the accuracy by 2%. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|