An Improved YOLO11 with Multi-Scale Feature Sequence Fusion for Corn Disease and Pest Detection

Reliable identification of corn diseases and insect pests is vital to precision agriculture; however, accurate detection in field environments remains difficult due to the small size of targets and complex backgrounds. To overcome these limitations, this study presents an enhanced object detection framework, built on YOLO11, for corn disease and pest recognition. An additional low-level feature layer (P2) is incorporated to retain fine-grained spatial details, thereby strengthening the representation of small objects. Furthermore, a multi-scale feature sequence fusion module is introduced to capture inter-scale correlations by aggregating features along the scale dimension, which effectively enriches contextual information across different resolutions. To further improve feature integration, a scale-aligned aggregation strategy is employed to balance semantic and spatial cues from multiple network depths. Experimental results on a self-collected corn disease and pest dataset show that the proposed method achieves an mAP@0.5 of 0.928, exceeding the baseline YOLO11 model by 3.3 %, while preserving real-time inference capability. These findings demonstrate the effectiveness and applicability of the proposed framework for practical corn disease and pest monitoring scenarios.