In Southeast Asia, people use sweet potato leaves as a type of vegetable for cooking, and they are also used as animal feed. Sweet potato leaves are commonly used in traditional medicine to treat tumors in the mouth and throat. There are different types of sweet potato leaves, including antin sweet potato leaves, papua salosa sweet potato leaves, sweet potato leaves, and Cilembu sweet potato leaves. The method used in this study is to detect the quality of sweet potato leaves, namely by going through the dataset process, bounding box, then the training process and finally the evaluation using the Mean Average Precision (mAP).
The results of the quality detection test for sweet potato leaves using YOLOv4-Tiny can be concluded that the maximum results according to the grade of each sweet potato leaf starting from the antin yam leaves have an MAP result of 0.79 on grade_a. Furthermore, sweet potato leaves with grade_b have an MAP value of 0.67. The papua salosa type of sweet potato leaves grade_a has a value of 0.70. The process of testing sweet potato leaves on grade_a has an MAP result of 0.82 on grade_b. The last test of Cilembu sweet potato leaves has a mean average precision (mAP) value at grade_a, which has a value of 0.75, and for grade_b. It has a mean average precision result of 0.40. As a result, it can be inferred that the utilization of YOLOv4-Tiny for identifying the quality of sweet potato leaves has been effective.