Classification of Vegetable Images Using Texture and Color Features

In this thesis, the aim to use vegetable images and implement a computationally cheap system to automatically classify vegetables using their texture and color features. In this respect, Scale Invariant Feature Transform (SIFT) and Speeded Up Robust Features (SURF) approaches are used to classify vegetable features. Feature extraction is done based on three color space channels; XYZ color space, HSV color space and RGB color space. It generates the features using color space channels. The classifier is then utilized once the vegetable features have been created for each image. Experiments are conducted on Kaggle Vegetable Image Dataset using 15 different varieties of popular vegetables found all over the world that include bean, bitter gourd, bottle gourd, brinjal, broccoli, cabbage, capsicum, carrot, cauliflower, cucumber, papaya, potato, pumpkin, radish and tomato, and the results will be presented at the end of the thesis. Comparison of the effect of SIFT and SURF methods on different color space channels for vegetable classification is demonstrated.