To address the challenges of corn-associated weed detection, such as diverse shapes, dense occlusion, complex backgrounds and scale variation, an improved object detection model, WAAP-YOLO, was proposed. First, the backbone was improved by replacing some convolutions with wavelet pooling convolutions, effectively avoiding aliasing artifacts. Second, an aggregated attention mechanism was introduced to construct the C2f-AA module, improving the model′s ability to extract weed features in complex backgrounds. Finally, ASF-P2-Net was proposed to replace the original neck network, incorporating the P2 detection head through the scale sequence fusion module, reducing model complexity and significantly improving small object detection performance. Experimental results show that the WAAP-YOLO detection algorithm achieves 97.2% mAP@0.5, 85.8% mAP@0.5∶0.95, 94.0% F1 score, and a parameter count of 2.1×106, outperforming common object detection models such as YOLOv5s, YOLOv8n, and YOLOv10n. The proposed model can significantly enhance cornfield weed recognition accuracy, which provides some reference for advancing the intelligent and sustainable development of the agricultural industry.