To counter the defect of traditional genetic algorithms, an improved adaptivegenetic algorithm with the criterion of premature convergence is provided. The occurrence ofpremature convergence is forecasted using colony...To counter the defect of traditional genetic algorithms, an improved adaptivegenetic algorithm with the criterion of premature convergence is provided. The occurrence ofpremature convergence is forecasted using colony entropy and colony variance. When prematureconvergence occurs, new individuals are generated in proper scale randomly based on superiorindividuals in the colony. We use these new individuals to replace some individuals in the oldcolony. The updated individuals account for 30 percent - 40 percent of all individuals and the sizeof scale is related to the distribution of the extreme value of the target function. Simulationtests show that there is much improvement in the speed of convergence and the probability of globalconvergence.展开更多
针对田间密植环境棉花精准打顶时,棉花顶芽因其小体积特性所带来识别困难问题,该研究提出一种改进型快速区域卷积神经网络(Faster Region Convolutional Neural Networks,Faster R-CNN)目标检测算法实现大田环境棉花顶芽识别。以Faster ...针对田间密植环境棉花精准打顶时,棉花顶芽因其小体积特性所带来识别困难问题,该研究提出一种改进型快速区域卷积神经网络(Faster Region Convolutional Neural Networks,Faster R-CNN)目标检测算法实现大田环境棉花顶芽识别。以Faster R-CNN为基础框架,使用RegNetX-6.4GF作为主干网络,以提高图像特征获取性能。将特征金字塔网络(Feature Pyramid Network,FPN)和导向锚框定位(Guided Anchoring,GA)机制相融合,实现锚框(Anchor)动态自适应生成。通过融合动态区域卷积神经网络(Dynamic Region Convolutional Neural Networks,Dynamic R-CNN),实现训练阶段检测模型自适应候选区域(Proposal)分布的动态变化。最后在目标候选区域(Region of Interest,ROI)中引入目标候选区域提取器(Generic ROI Extractor,GROIE)提高图像特征融合能力。采集自然环境下7种不同棉花总计4819张图片,建立微软常见物体图像识别库2017(Microsoft Common Objects in Context 2017,MS COCO 2017)格式的棉花顶芽图片数据集进行试验。结果表明,该研究提出方法的平均准确率均值(Mean Average Precision,MAP)为98.1%,模型的处理帧速(Frames Per Second,FPS)为10.3帧/s。其MAP在交并比(Intersection Over Union,IOU)为0.5时较Faster R-CNN、RetinaNet、Cascade R-CNN和RepPoints网络分别提高7.3%、78.9%、10.1%和8.3%。该研究算法在田间对于棉花顶芽识别具有较高的鲁棒性和精确度,为棉花精准打顶作业奠定基础。展开更多
基金The Natural Science Foundation of Jiangsu Province (BK99011).
文摘To counter the defect of traditional genetic algorithms, an improved adaptivegenetic algorithm with the criterion of premature convergence is provided. The occurrence ofpremature convergence is forecasted using colony entropy and colony variance. When prematureconvergence occurs, new individuals are generated in proper scale randomly based on superiorindividuals in the colony. We use these new individuals to replace some individuals in the oldcolony. The updated individuals account for 30 percent - 40 percent of all individuals and the sizeof scale is related to the distribution of the extreme value of the target function. Simulationtests show that there is much improvement in the speed of convergence and the probability of globalconvergence.
文摘针对田间密植环境棉花精准打顶时,棉花顶芽因其小体积特性所带来识别困难问题,该研究提出一种改进型快速区域卷积神经网络(Faster Region Convolutional Neural Networks,Faster R-CNN)目标检测算法实现大田环境棉花顶芽识别。以Faster R-CNN为基础框架,使用RegNetX-6.4GF作为主干网络,以提高图像特征获取性能。将特征金字塔网络(Feature Pyramid Network,FPN)和导向锚框定位(Guided Anchoring,GA)机制相融合,实现锚框(Anchor)动态自适应生成。通过融合动态区域卷积神经网络(Dynamic Region Convolutional Neural Networks,Dynamic R-CNN),实现训练阶段检测模型自适应候选区域(Proposal)分布的动态变化。最后在目标候选区域(Region of Interest,ROI)中引入目标候选区域提取器(Generic ROI Extractor,GROIE)提高图像特征融合能力。采集自然环境下7种不同棉花总计4819张图片,建立微软常见物体图像识别库2017(Microsoft Common Objects in Context 2017,MS COCO 2017)格式的棉花顶芽图片数据集进行试验。结果表明,该研究提出方法的平均准确率均值(Mean Average Precision,MAP)为98.1%,模型的处理帧速(Frames Per Second,FPS)为10.3帧/s。其MAP在交并比(Intersection Over Union,IOU)为0.5时较Faster R-CNN、RetinaNet、Cascade R-CNN和RepPoints网络分别提高7.3%、78.9%、10.1%和8.3%。该研究算法在田间对于棉花顶芽识别具有较高的鲁棒性和精确度,为棉花精准打顶作业奠定基础。
