【摘要】 番茄目标检测在自动化番茄采摘工作中发挥巨大作用，为了提高番茄目标检测模型的识别精度和速度，提出了一种基于改进YOLOv11的轻量化番茄采摘目标检测方法。YOLOv11模型引入了新的C3K2和C2PSA模块，改进了检测头，以提高模型特征提取能力和减少模型计算量。改进后的YOLOv11网络结构借鉴ShuffleNet-v2网络优点，用Shuffle Block模块替换Backbone中的CBS模块，减少计算参数量，以提高模型的检测速度。实验结果表明，改进后模型相比于原模型，在实际的番茄检测中，仅损失了0.72%的准确率和1.65%的召回率，FPS提高了11.6%，表现出优异的检测性能。最后，在实机上进行模拟采摘测试，验证了该模型在番茄采摘工作中应用效果较好。更多还原

【Abstract】 Tomato object detection plays a significant role in automated tomato harvesting. To enhance detection accuracy and speed, this paper proposes a lightweight tomato harvesting object detection method based on an improved YOLOv11. The YOLOv11 model introduces new C3K2 and C2 PSA modules, along with an improved detection head, to enhance feature extraction capability and reduce computational load. The modified YOLOv11 network structure incorporates advantages from the ShuffleNet-v2 network, replacing the CBS module in the backbone with a Shuffle Block module, effectively reducing parameters to boost detection speed. Experimental results show that, compared to the original model, the improved model achieves a detection performance with only a 0.72% loss in accuracy and a 1.65% reduction in recall,while FPS increases by 11.6%. Finally, a simulated harvesting test conducted on real machines demonstrates the model’s practical applicability in tomato harvesting tasks.更多还原