Automatic Recognition for Tomatoes Based on Computer Vision Algorithm

Abstract

The lack of use of computer vision technology to detect ripeness in sorting tomatoes has
made farmers still use conventional methods, namely sorting based on direct visual
observation of the fruit to be sorted. The disadvantage of categorizing it visually is direct
which is very subjective, so it is inconsistent with the sorting process for the level of ripeness
of the tomatoes. Therefore, sorting tomatoes utilizing technological sophistication needs to be
done. The construction of this system uses a Radial Basic Neural Network Algorithm with
RGB value parameters, and the mean, entropy, and variance values using a Gabor filter in
the form of GUI (Graphical User Interface) connected to a webcam equipped with a servo
which plays an important role as a tomato fruit sorting mechanism. This is expected to be
able to help farmers in the sorting process for ripe tomatoes. In this experiment, a sorting
system was carried out to detect maturity based on color changes in objects, in the
experiment an accuracy was obtained that the computer vision algorithm was able to
recognize and sort tomatoes by up to 90% using random sample data.