In the process of smart orchard construction, accurate detection of target fruit is an important guarantee to realize intelligent management of orchards. Green apple detection technology greatly diminishes the need for manual labor, cutting costs and time, while enhancing the automation and efficiency of sorting processes. However, due to the complex orchard environment, the ever-changing posture of the target fruit, and the difficulty of detecting green target fruit similar to the background, they bring new challenges to the detection of green target fruit. Aiming at the problems existing in green apple detection, this study takes green apples as the research object, and proposes a green apple detection model based on optimized deformable DETR. The new method first introduces the ResNeXt network to extract image features to reduce information loss in the feature extraction process; secondly, it improves the accuracy and optimizes the detection results through the deformable attention mechanism; and finally, it uses a feed-forward network to predict the detection results. The experimental results show that the accuracy of the improved detection model has been significantly improved, with an overall AP of 54.1, AP₅₀ of 80.4, AP₇₅ of 58.0, AP_s of 35.4 for small objects, AP_m of 60.2 for medium objects, and AP_l of 85.0 for large objects. It can provide a theoretical reference for green target detection of other fruit and vegetables green target detection.

Accurate fruit detection in complex orchard environments remains challenging due to variable lighting conditions and weather factors. This paper proposes an optimized contour segmentation model for green spherical fruits (apples and persimmons) based on the E2EC framework. The model employs DLA34 as the backbone network for feature extraction enhanced by a path aggregation balanced feature pyramid network (PAB FPN) with embedded attention mechanisms to refine feature representation. For contour segmentation, we introduce a Cycle MLP Aggregation Deformation (CMAD) module that incorporates cycleMLP to expand the receptive field and improve contour accuracy. Experimental results demonstrate the model’s effectiveness, achieving average precision (AP) and average recall (AR) of 75.5% and 80.4%, respectively, for green persimmons and 57.8% and 64.0% for green apples, outperforming previous segmentation methods. These advancements contribute to the development of more robust smart agriculture systems.