Tree Seedlings Detection and Counting Using a Deep Learning Algorithm

Tree-counting methods based on computer vision technologies are low-cost and efficient in contrast to the traditional tree counting methods, which time-consuming, laborious, humanly infeasible. This study presents a method for detecting seedlings images using deep learning algorithm with high economic value broad application prospects type quantity of seedlings. The dataset was built three types seedlings: dragon spruce, black chokeberries, Scots pine. data were augmented via several augmentation improve accuracy detection model prevent overfitting. Then YOLOv5 object network trained obtain training weights. results experiments showed that our proposed could effectively identify count an image. Specifically, MAP pine 89.8%, 89.1%, 95.6%, respectively. reached 95.10% average (98.58% 91.62% 95.11% pine). can provide technical support statistical tasks trees.