EPSA-YOLO-V5s: A novel method for detecting the survival rate of rapeseed in a plant factory based on multiple guarantee mechanisms

Abstract

• The dataset of key growth stages of rape under the plant factory is obtained. • We established a target detection model based on multiple key growth stages. • A triple guarantee mechanism was proposed for the crop survival rate detection. As one of the important products of modern agricultural development, plant factories can provide a suitable environment for the growth and development of crops. Intelligently detecting the survival rate of crops in multiple key growth stages can not only improve the space utilization of plant factory, but also help increase crop yields. In this work, our main task is to use a novel method to detect the survival rate of rape seedlings at multiple growth stages in the plant factory. First of all, for the key growth stages where seedlings may die, we obtained image datasets of the whole process of seed germination, the early, and the middle stage of seedling transplanting. Second, we used the state-of-the-art method YOLO-V5s to construct the target detection model for the rape seedling dataset of the three key growth stages, and achieved good performance of the model mAP@0.5 as 0.994, 0.996, and 0.996 respectively. Finally, in order to construct a model suitable for the detection of the survival rate of rape in multiple key growth stages, we propose a new method called ESPA-YOLO-V5s, and achieved a good model performance with a mAP@0.5 of 0.996. The experimental results prove that our method has laid a good foundation for the survival rate detection of the key growth stages of plant.