Crop Node Detection and Internode Length Estimation Using an Improved YOLOv5 Model

The extraction and analysis of plant phenotypic characteristics are critical issues for many precision agriculture applications. An improved YOLOv5 model was proposed in this study accurate node detection internode length estimation crops by using an end-to-end approach. In YOLOv5, a feature module added front each head, the bounding box loss function used original network replaced SIoU function. results experiments on three different (chili, eggplant, tomato) showed that reached 90.5% AP (average precision) average time 0.019 s per image. error 41.3 pixels, relative 7.36%. Compared with had reduction 5.84 pixels 1.61%.