발간년도 : [2021]
| 논문정보 |
|
| 논문명(한글) |
[Vol.16, No.6] A Study on the YOLO-based Moth Larva Detection Model |
|
| 논문투고자 |
Si-Young Rho, Kang-Su Kwak |
|
| 논문내용 |
Moth pests are pests that cause a lot of damage to food crops. Moth larvae are similar in shape and shape. In addition, even moth larvae have different colors depending on the environment and consummation, so it is difficult for skilled farmers to accurately identify moth larvae. However, accurate identification of moth larvae is very important because moth larvae have different control methods. The main user of the moth larva detection model is a farmer without computer knowledge. Thus, the most important factor is not the time or accuracy to detect larvae of moths. How little false positive or false negative is a more important factor. It prevents waste of labor and financial losses of farmers who have misrepresented by error detection. This is because it can improve the credibility of smart agricultural technology. The YOLO v2, YOLO v3, and YOLO v4 models were learned as the same moth larvae image and the performance between models was compared. As a result of the experiment, the YOLO v4model showed Precision, Recall, and F1-Score as 1.00. In addition, it was confirmed that the performance was superior to other YOLO models with 88.39% of IoU and 79.96% of mAP. False Positive or False Negative is also less than other YOLO models. It was confirmed that the YOLO v4model was suitable as a moth larva detection model. |
|
| 첨부논문 |
|
|
|
|
|
자료실 > 논문지 
