Abstract: To address the insufficient fine-grained classification capability and suboptimal localization accuracy in current ship target recognition tasks, this paper proposes an improved YOLO-based algorithm named fine-grained classification and precise positioning YOLO (FP-YOLO). The algorithm introduces novel cross stage partial bottleneck modules (D2f and G2f) into the backbone and neck networks, respectively, enhancing both the backbone's detail feature extraction capability and the neck's global information interaction capacity. Furthermore, a backbone-neck feature fusion module (BNFM) is designed to effectively integrate the same-level feature maps, optimizing the complementary characteristics between the backbone and neck features while improving the neural network's generalization ability. Additionally, a sliding loss function is introduced to better utilize hard-classified samples during training. Experimental results demonstrate that the proposed algorithm achieves 66.9% mAP@0.95 on a custom-built horizontal fine-grained ship dataset, outperforming YOLOv8n and YOLOv10n by 2.5 and 5.4 percentage points, respectively. On a public remote sensing fine-grained ship dataset, it reaches 57.2% mAP@0.5:0.95, improving by 3.4 and 7.0 percentage points compared to the above two models.