Abstract: Real-time monitoring of flat peaches across different growth stages in natural orchard environments is crucial for yield estimation, precise fertilization, and automated harvesting. However, traditional methods can only detect peaches at a single specific stage, exhibiting poor adaptability across different growth phases of the same fruit. Severe occlusion from peach branches and foliage, combined with dense overlapping clusters of young fruit and significant variations in size, color, fuzziness and texture during growth, makes accurate identification particularly challenging during the color change and enlargement stages. To address the insufficient accuracy in identifying flat peach growth stages within natural orchard environments, this paper proposed an Orchard Growth Detection Network (OGD-Net) based on the lightweight YOLOv11n model. First, an improved adaptive orchard downsampling module (AOD-Module) was designed by integrating the context guided block (CG Block) with depthwise separable convolution to replace the original downsampling layer in the backbone network, enhancing peach feature expression while reducing computational cost. Second, an orchard gated spectrum block (OGS-Block) was proposed by combining frequency-domain dynamic convolution with a gating mechanism, which replaced the original C3K2 module to enable dual frequency-spatial perception and enhance precise identification across different growth stages. Third, an orchard-adaptive feature fusion pyramid (OAFF-Pyramid) was constructed in the neck network. Within OAFF-Pyramid, an orchard attentive upsampler (OAS-Sampler) was designed by combining EUCB and the Shift Channel Mix mechanism to reconstruct fruit contours occluded by branches and leaves; an OAF-Block was designed by fusing C2f with MSCB, CSP, and Shift Channel Mix to replace the original C3k2 feature extraction module. Finally, a separated and enhancement attention module (SEAM) was introduced at the detection head to compensate for feature loss caused by occlusion and improve model robustness under branch and leaf occlusion conditions. A flat peach dataset was constructed by collecting 4 807 original images at the 143rd Regiment flat peach orchard in Shihezi, Xinjiang, China, covering three growth stages, namely the green and hard stage, the expansion and color transition stage, and the mature stage. After data augmentation applied only to the training set, a final dataset of 11 535 images was obtained. Experimental results demonstrated the following: 1) The OGD-Net model achieved 6.0 G floating-point operations and a model size of 4.7 MB. On the test dataset spanning three different growth stages of flat peaches, the model achieved precision, recall, F₁ score, mAP_0.5, mAP_0.75, and mAP_0.5–0.95 of 88.17%, 84.60%, 86.31%, 93.14%, 89.12%, and 85.48%, respectively, representing improvements of 0.75, 3.25, 2.04, 3.28, 5.36, and 6.38 percentage points over YOLOv11n. 2) Compared with YOLOv11n, the floating-point computation volume and model size of OGD-Net were reduced by 4.76% and 9.62%, respectively, indicating excellent lightweight characteristics suitable for mobile-terminal deployment. 3) For the challenging fruit enlargement and color transition stage, the detection accuracy mAP0.5 reached 89.63%, representing a 4.51 percentage point improvement over YOLOv11n. 4) Compared with classic neck network architectures such as BiFPN, GFPN, Gold-YOLO, and CGRFPN, the proposed OAFF-Pyramid achieved an R-value of 83.97% and mAP0.5 of 90.72%, with a model size of only 4.5 MB, which was 13.46% smaller than the baseline. The heatmap highlights were highly concentrated on key characteristic areas of peaches, with no false detections or background false responses observed, demonstrating that the OAFF-Pyramid architecture exhibited superior detection capabilities and stronger generalization performance for growing peaches in complex orchard environments. 5) To validate the practicality of the model, OGD-Net was deployed on an MMX5 Plus robot. Based on an indoor peach orchard simulation platform, it achieved recognition success rates of 100%, 100%, and 93.3% under no occlusion, branch and leaf occlusion, and fruit overlap conditions, respectively, and grasping success rates of 93.3%, 86.7%, and 80% for mature peaches under the corresponding conditions. The results demonstrated that the proposed OGD-Net model achieved accurate identification of flat peaches across the entire growth period in natural orchard environments and provided technical support for precise growth monitoring and automated harvesting operations during orchard production.