Target Detection Of Duck Based On YOLOv3 Model

Aiming at the scientific management of duck breeding, the method of duck target detection was studied by taking the living conditions of duck in the actual breeding environment as the experimental sample data set. In order to reduce the cost of manual counting, a duck target detection model is built based on the depth neural network's YOLOv3[1] algorithm. The model is tested on the self-made data set, and the recall rate and precision are greater than 0.65, which has achieved good results.