Aiming at the requirement of both high detection accuracy and high detection speed in the object detection fieldBalanced YOLO (B-YOLO)a one-stage object detection algorithm is proposed.Firsta spatial attention mechanism is introduced in the algorithmand the multi-scale maximum pooling layer is used to increase the range of the receptive field.Thenthe cross-stage local connection structure and the pass-through layer are adopted to optimize the backbone network structure and improve the computational efficiency.Finallya bottom-to-top path is added to the multi-scale detection structureand splicing is used for transverse connection to effectively fuse the deep semantic information with the shallow location information. Experimental results show that the algorithm has achieved a good balance between accuracy and speed.