In this paper, we propose a different method of deepening the network model of YOLOv5s6 and design three types of Multi-Connection (MC) blocks that are suitable for specific datasets. The main purpose of Multi-Connection block is to reuse features and retain input features for passing down. Eight public datasets and one customized dataset are experimented for verification. We improve the residual block in YOLOv5. The results show that the average precision (AP) can be increased. Compared with the YOLOv5s6, YOLOv5s6 with MC type I increases the AP by 1.6% in the Global Wheat Head Dataset 2020, YOLOv5s6 with MC type II increases the AP by 2.9% in the PlanDoc dataset, and YOLOv5s6 with MC type III increases the AP by 2.9% in the PASCAL Visual Object Classes (VOC) dataset. Using the multi-connection of double residual block performs better than the original residual block of YOLOv5s6.