SPD-YOLOv7: A New Method for Corn Pest Detection

Aiming at the problems of small pests, blurred image, low resolution and large species difference in different growth periods of corn crops. In this paper, an accurate and efficient method for the detection of maize crop pests, SPD-YOLOv7, was proposed. In this paper, target background analysis, data enhancement, Gaussian noise and brightness processing are used to improve the generalization and robustness of the model. In this paper, SPD-Conv module is introduced into the framework of SPD-YOLOv7 to replace part of the step convolutional layer in the traditional system backbone and head network, so as to realize the prominent feature and location information of small target. In addition, the ELAN-W module is combined with the CBAM attention mechanism to extract more efficient features. The accuracy of the improved YOLOv7 network model reaches 98.38%, and the average accuracy reaches 99.4%.