doTraining=false；outputFolder=fullfile（tempdir，“潘达塞特”）;lidarURL = ['https://ssd.mathworks.com/金宝appsupportfiles/lidar/data/'......'pandaset_lidardata.tar.gz'];HelperdownloadPandasetData（outputFolder，Lidarurl）;

路径= fullfile (outputFolder,激光雷达的）;lidarData = fileDatastore(路径,“ReadFcn”，@（x）pcread（x））;

gtPath = fullfile (outputFolder,'长方体','pandasetlidargroundtruthmat'）;data =负载(gtPath'lidargtlabels')；Labels=timetable2table（data.lidarGtLabels）；boxLabels=Labels（：，2:3）；

图ptCld=读取（lidarData）；ax=pcshow（ptCld.位置）；设置（ax，“XLim”50 [-50],“YLim”，[ -  40 40]）;变焦（斧头，2.5）;轴从；

重置(lidarData);

xMin=0.0；沿x轴的％最小值。ymin = -39.68;沿y轴的％最小值。zmin = -5.0;%沿Z轴的最小值。xMax = 69.12;沿x轴的最大值％。yMax=39.68；% y轴上的最大值。zMax = 5.0;沿z轴的％最大值。xStep = 0.16;沿x轴的分辨率。yStep = 0.16;沿y轴的分辨率。dsFactor=2.0；%将采样因素。％计算伪图像的尺寸。xn = round（（（xmax  -  xmin）/ xstep））;yn = round（（（ymax  -  ymin）/ ystep））;定义点云参数。yMin pointCloudRange = [xMin xMax, yMax, zMin, zMax);voxelSize = [xStep, yStep];

[croppedPointCloudObj, processedLabels] = cropFrontViewFromLidarData (......lidarData、boxLabels pointCloudRange);

pc=croppedPointCloudObj{1,1}；gtLabelsCar=processedLabels.Car{1}；gtLabelsTruck=processedLabels.Truck{1}；HelperDisplay3BoxesOverlaidPointCloud（pc.Location，gtLabelsCar，......“绿色”gtLabelsTruck,“红色”,“裁剪点云”）;

重置(lidarData);

rng（1）；ShuffledDices=randperm（大小（processedLabels，1））；idx=floor（0.7*长度（ShuffledDices））；trainData=croppedPointCloudObj（ShuffledDices（1:idx），：）；testData=croppedPointCloudObj（ShuffledDices（idx+1:end），：）；trainLabels=processedLabels（ShuffledDices:1:idx），：）；testLabels=processedLabels=processedLabels（idx+1:end），：）；

writefiles = true;datalocation = fullfile（outputfolder，“InputData”)；[trainData，trainLabels]=saveptCldToPCD（trainData，trainLabels，......dataLocation, writefile);

lds=文件数据存储（数据位置，“ReadFcn”，@（x）pcread（x））;

BDS = BoxLabeldAtastore（TrainLabels）;

cd =结合(摩门教,bds);

augData=read（cds）；augptCld=augData{1,1}；augLabels=augData{1,2}；augClass=augData{1,3}；labelsCar=augLabels（augClass==“汽车”:);Labelstruck =奥古拉标签（奥格拉斯=='卡车':);Helperdisplay3dboxesoverlaidPointCloud（奥古斯特·普利.Location，Labelscar，“绿色”,......标签车，“红色”,“数据扩充前”）;

重置（CD）；

一会= {“汽车”,'卡车'}；sampleLocation=fullfile（tempdir，“GTsamples”)[SamAssigntData，Indexs]=来自LidarData的SampleGroundTruthObject（CD，类名，......'minpoints', 20岁,“sampleLocation”, sampleLocation);

numObjects = (10,10);addsaugmented = transform(cd，@(x) augmentGroundTruthObjectsToLidarData(x，......sampledGTData指数,一会,numObjects));

cdsAugmented = transform(cdsAugmented，@(x) augmentData(x));

augData =阅读(cdsAugmented);augptCld = augData {1};augLabels = augData {1,2};augClass = augData {1,3};labelsCar = augLabels (augClass = =“汽车”:);Labelstruck =奥古拉标签（奥格拉斯=='卡车':);Helperdisplay3dboxesoverlaidPointCloud（奥古斯特·普利.Location，Labelscar，“绿色”,......标签车，“红色”,'数据增强后'）;

重置（CDS-ummented）;

％定义突出的支柱数量。P = 12000;％定义每支柱的点数。n = 100;从训练数据中估计锚盒。anchorBoxes = calculateAnchorsPointPillars (trainLabels);一会= trainLabels.Properties.VariableNames;定义pointcolumns检测器。探测器= PointpillarsObjectDetector（Pointcroudrange，ClassNames，Chandorboxes，......'voxelsize'，毒素，“NumPillars”，p，“NumPointsPerPillar”，n）;

执行环境=“汽车”；如果CanUseCallallPool DispartinBackground = True;其他的dispatchInBackground = false;终止选择= trainingOptions ('亚当',......“情节”,“没有”,......“MaxEpochs”,60,......'minibatchsize'，3,......“梯度衰减因子”, 0.9,......“SquaredGradientDecayFactor”,0.999,......'shownrateschedule',“分段”,......“初始学习率”,0.0002,......“LearnRateDropPeriod”15，......“LearnRateDropFactor”,0.8,......“执行环境”executionEnvironment,......“DispatchInBackground”，调度背景，......“BatchNormalizationStatistics”,'移动',......“ResetInputNormalization”假的,......“CheckpointPath”, tempdir);

如果doTraining[detector，info]=TrainPointPillarObjectDetector（CDS分段，检测器，选项）；其他的pretrainedDetector =负载(“预训练点柱检测器。垫”,“探测器”）;探测器= pretroseddetector.detector;终止

ptCloud=testData{45,1}；gtLabels=testLabels（45，：）；%指定仅使用的检测的置信阈值%置信度得分高于此值。信心阈值=0.5；[框、分数、标签]=检测（检测器、云端、，'临界点', confidenceThreshold);boxlabelsCar(标签' = = =框“汽车”:);boxlabelsTruck(标签' = = =框'卡车':);%在点云上显示预测。helperDisplay3DBoxesOverlaidPointCloud (ptCloud。位置、boxlabelsCar“绿色”,......Boxlabelstruck，“红色”,“预测边界框”）;

numinputs = 50;％从长方体标签生成旋转矩形。bds = boxLabelDatastore (testLabels (1: numInputs,:));groundTruthData = transform(bds，@(x) createRotRect(x));%设置阈值。NMSPositiveOutThreshold=0.5；置信阈值=0.25；检测结果=检测（检测器，测试数据（1:numInputs，：），......'临界点', confidenceThreshold);%转换为旋转矩形格式以计算度量对于i = 1:height(detectionResults) box = detectionResults. boxes {i};detectionResults。盒子我{}=盒(:,(1、2、4、5、7]);终止度量=评估的检测AOS（检测结果、地面数据、，......nmspositiveiouthreshold）;DISP（指标（：，1：2））

函数HelperdownloadPandasetdata（outputfolder，lidarurl）%将数据集从给定URL下载到输出文件夹。lidarDataTarFile = fullfile (outputFolder,'pandaset_lidardata.tar.gz'）;如果~存在(lidarDataTarFile'文件'mkdir (outputFolder);disp ('下载Pandaset Lidar驾驶数据（5.2 GB）......'）;websave (lidarDataTarFile lidarURL);解压(lidarDataTarFile outputFolder);终止%解压文件。如果(~存在(fullfile (outputFolder激光雷达的),“dir”))......& & (~ (fullfile (outputFolder,存在'长方体'),“dir”)解压(lidarDataTarFile outputFolder);终止终止函数helperDisplay3DBoxesOverlaidPointCloud (ptCld labelsCar carColor,......labelsTruck、truckColor titleForFigure)%显示具有不同颜色边界框的点云，以显示不同颜色的点云%的类。数字;AX = PCSHOW（PTCLD）;showshape（'长方体'，labelscar，“父”，斧头，“不透明度”,0.1,......“颜色”carColor,'行宽'，0.5）；保持在；showshape（'长方体'labelsTruck,“父”，斧头，“不透明度”,0.1,......“颜色”truckColor,'行宽', 0.5);标题(titleForFigure);变焦(ax, 1.5);终止