| dc.description.abstract | In industrial development nowadays, chili as one of the staples is needed in the
production process. To obtain the desired level of chili ripeness, sorting the level of
ripeness of red chili is needed so that it can meet market needs and industrial needs.
The ripeness level is one of the most important factors to consider during the red pepper
sorting process. The sorting process in determining the level of ripeness of red chili is
usually done visually and manually directly on the red chili fruit. Some of the
weaknesses in the process of determining this maturity level are different levels of
perceived ripeness, and require a lot of labor to sort. Therefore, a system is needed that
is able to detect the level of doneness of red chili accurately in real-time. To achieve
this, this study used the YOLOv4 method to detect the ripeness level of ripe, half-ripe,
and raw red chili based on Android. YOLOv4 uses CSPDarknet53 which can improve
CNN learning ability, eliminate computational bottlenecks, and reduce memory loss.
From the research conducted using 1200 training data and 300 testing data, it got 100%
accuracy with a total of 1500 red chili images. | en_US |
