Distinguishing of Rice Varieties by Using Machine Learning Models

Abstract

A large number of studies have been executed for classifying plant types and identifying diseases of various crops particularly using images. The plant type identification problem is further complicated by common object recognition difficulties mainly due to light, pose and orientation. The present study was undertaken to distinguish the four different Indian rice varieties by utilizing the respective collected features and applying machine learning methods. There are various methods available from simple to complex model. However, the study was carried out with simple models like Linear Discriminant Analysis, Logistic Regression, K-Nearest Neighbours and Naïve-Bayes method. K-NN out performed over the other methods with an Accuracy and Precision of 99.16% and 99% respectively. CONCLUSIONIn this study, previous work of different approaches was presented by their pros and cons. The results obtained based on employing feature normalization and K-NN model are quite promising, which suggest that this method can provide an accurate solution to the rice varieties for their distinguishing and/or identification problem alternative to sophisticated image segmentation techniques as discussed earlier. Future work is being more focused towards our selfcollected dataset to include more rice varieties to address automatic identification of rice varieties in particular and other field crops varieties in general.

Identification of Rice Quality Through Pattern Classification Using Computer Vision Image Processing

“…Different methods used such as App List, K-Neighbor Neighbor's and Naïve-Bayes. KNN implemented to other methods with accuracy, accuracy and recall of 99.16%, 99.12% and 99.12% respectively [6].…”

“…However, the study was conducted on specific models, such as discriminant analysis, logistic disturbances, K-Neighbors and Naive-Bayes neighbors. The index was made using other methods with an accuracy and accuracy of 99.16% and 99%, respectively [20].…”

“…However, having a large number of different types makes it difficult to test and classify as a new employee. Variations can be distinguished in the field, determining their height, the location of the decay, the type of seed and the color, and so on [6].…”

A visual dataset for recognition of rice varieties

“…There are numerous varieties (approx. 40,000) of rice, each exhibiting unique characteristics in terms of size, shape, color, and aroma [2] . Popular varieties include Basmati, Jasmine, Arborio, and the various types of short-grain and long-grain rice.…”