As an important ornamental plant, the automatic detection and classification of the maturity of Alstroemeria Genus Morado flowers hold significant importance in precision agriculture. However, this task faces numerous challenges due to the diversity of morphological characteristics, complex growth environments, and factors such as occlusion and lighting variations. Currently, this field is relatively unexplored, necessitating innovative methods to overcome existing difficulties. To fill this research gap, this study developed a deep learning-based object detection framework, the Alstroemeria Genus Morado Network (AGMNet), specifically optimized for the detection and classification of Alstroemeria Genus Morado flowers. This convolutional neural network utilizes multi-scale feature fusion techniques and spatial attention mechanisms, along with a dual-path detection structure, significantly enhancing its capability for automatic maturity classification and detection of flowers. Notably, AGMNet addresses the issue of class imbalance in its design and employs advanced data augmentation techniques to enhance the model's generalization ability. In comparative experiments on the morado_5may dataset, AGMNet demonstrated superior performance in Precision, Recall, and F1-score, with a 3.8% improvement in the mAP metric over the latest YOLOv9 model, showcasing stronger generalization capabilities. AGMNet is expected to play a more significant role in enhancing agricultural production efficiency and automation levels.