DoR

Scopus Indexed Publications

Abstract: We often travel to different countries, but we are not aware of the local vegetables of many countries. If we are aware of the local vegetables of different countries, then it is very easy for us to recognize the specific local vegetables. Since it is very important to recognize the local vegetables in our daily life. But it's challenging to identify easily & within short time. Because most of the vegetables are almost similar color, similar size and similar shapes. So through my paper I have shown the recognition process of five (5) types of local vegetables. Which will be helpful to our farmer as well as general people. In this work I proposed some machine learning methods with computer vision approach for local vegetables recognition and identification. Segmentation of local vegetable pixels is performed using K-means clustering, SVM classifier is used for image classification and feature is extracted according to two steps explicitly: Gray level Co-occurrence matrix and statistical features. For each local vegetables confusion matrix in binary format had been performed. My proposed system is much better than any other research work. Because my paper have achieved 97.97% average accuracy. Where the individual highest accuracy is 98.20% (Bottle Ground) and lowest accuracy is 97.69% (Potato). It also shows the better performance in sensitivity, precision and specificity where their values are respectively 94.90%, 94.37% and 98.77%.

Keywords: local vegetable , recognition, machine learning, computer vision, image processing, automation

Journal or Conference Name: 2021 IEEE 4th International Conference on Computing, Power and Communication Technologies (GUCON)