The inflorescence developmental stage of alfalfa serves as a critical physiological indicator determining its nutritional value and yield, making real-time and precise monitoring highly significant for forage quality regulation and cultivation management optimization. To address the technical challenges in large-scale alfalfa cultivation, including small inflorescence targets (8 ~ 32 pixels), dense distribution, complex backgrounds, and high demands for real-time monitoring, an intelligent monitoring method for alfalfa inflorescences was proposed by integrating low-altitude UAV image and lightweight deep learning. Firstly, a multi-scenario, multi-variety, and multi-temporal low-altitude UAV RGB dataset of alfalfa inflorescences was constructed, enhancing model generalizability through sample filtering, hybrid sample selection, and data augmentation strategies. Secondly, an improved YOLO 11n ICA model was designed. To effectively enhance the model's ability to detect dense small targets in complex backgrounds, the C3K2_INXB module was innovatively designed and the max-avg context anchor attention (MACAA) module was constructed. Finally, a real-time alfalfa inflorescence monitoring system was implemented based on a Web-cloud collaborative architecture, achieving an inference speed of 38 f/s. Experimental results demonstrated that the proposed method achieved a mean average precision (mAP@50) of 97.5% and a small-target recall rate of 92.9% on the self-built low-altitude alfalfa inflorescence dataset. Field validation tests showed that the system achieved a recognition accuracy of 95.28% and the miss-detection rate was 4.72%. Additionally, it automatically generated spatiotemporal distribution heatmaps of alfalfa inflorescences, providing decision-making support for precision agriculture management. The research result can provide a high-precision, lightweight, and real-time approach for rapid inflorescences counting in complex field environments, offering significant practical value for advancing intelligent pasture management.