A Machine Learning Based Automatic Tomato Classification System

In this paper, a machine learning based automatic tomato classification system is proposed to estimate the ripeness of the tomatoes. In the proposed method, a preprocessing step is first devised to captured a tomato from a given image. Then, Gist feature extraction algorithm is constructed to acquire the multidimensional features of tomato. For each dimensional feature, the corresponding estimation stage is derived from the output of Squeezenet model by minimizing the loss function. Finally, the final stage is determined via the voting results. The effectiveness and feasibility of the proposed system are verified on the some images captured from different environments.