ans = 23x1图层数组227 x227x3数据的图像输入图像的zerocenter正常化2 conv1卷积96年11 x11x3旋转步[4 4]和填充[0 0 0 0]3‘relu1 ReLU ReLU 4“norm1”横通道正常化横通道规范化5频道每个元素5“pool1”马克斯池3 x3 Max池步(2 - 2)和填充[0 0 0 0]6conv2卷积256 5 x5x48旋转步[1]和填充(2 2 2 2)7的relu2 ReLU ReLU 8 norm2的横通道正常化横通道正常化与5频道/元素9“pool2”马克斯池3 x3马克斯池步(2 - 2)和填充[0 0 0 0]10 conv3卷积384 3 x3x256[1]和隆起与进步填充[1 1 1 1]11的relu3 ReLU ReLU 12 conv4卷积384 3 x3x192旋转步[1]和填充[1 1 1 1]13的relu4 ReLU ReLU 14 conv5卷积256 3 x3x192旋转步[1]和填充[1 1 1 1]15 ' relu5 ReLU ReLU 16“pool5”马克斯池3 x3 Max池步(2 - 2)和填充[0 0 0 0]17 fc6完全Connected 4096 fully connected layer 18 'relu6' ReLU ReLU 19 'drop6' Dropout 50% dropout 20 'fcLane1' Fully Connected 16 fully connected layer 21 'fcLane1Relu' ReLU ReLU 22 'fcLane2' Fully Connected 6 fully connected layer 23 'output' Regression Output mean-squared-error with 'leftLane_a', 'leftLane_b', and 4 other responses

函数[laneFound，ltPts，rtPts]=从网络输出中检测车道（帧，laneCoeffMeans，laneCoeffStds）%1，计算%图像坐标中的左右车道点。摄像机坐标由加州理工学院%mono摄像机模型描述#codegen%mynet持久对象用于加载series网络对象。%在第一次调用此函数时，将构造持久对象并%setup。后续调用该函数时，将重用同一对象%以调用输入上的predict，从而避免重建和重新加载%网络对象。宿存花被；如果isempty（lanenet）lanenet=coder.loadDeepLearningNetwork（'lanenet.mat'，'lanenet'）；end lanecoeffsNetworkOutput=lanenet.predict（permute（frame[2 1 3]）；%通过反转规范化步骤params=lanecoeffsNetworkOutput.*laneCoeffStds+lanecoeffsmeans恢复原始系数；isRightLaneFound=abs（参数（6））>0.5；%c应大于0.5，以使其成为右车道isLeftLaneFound=abs（参数（3））>0.5；车辆积分=3:30；%米，传感器前面ltPts=编码器。零拷贝（零（28,2，'single'）；rtPts=编码器.nullcopy（零（28,2，'single'）；如果isRightLaneFound和isLeftLaneFound rtBoundary=params（4:6）；rt_y=计算边界模型（rtBoundary，车辆点）；ltBoundary=参数（1:3）；lt_y=计算边界模型（ltBoundary，车辆点）；%可视化ego车辆的车道边界tform=获取\u tformToImage；%将车辆映射到图像坐标ltPts=t形式转换点相反（[vehicleXPoints'，lt_y']）；rtPts=t形式转换点相反（[vehicleXPoints'，rt_y']）；laneFound=true；else-laneFound=false；结束函数yWorld=computeBoundaryModel（model，xWorld）yWorld=polyval（model，xWorld）；结束函数tform=get_tformToImage%基于相机设置计算外部偏航=0；螺距=14；%摄像机的俯仰角（以度为单位）滚动=0；平移=平移向量（偏航、俯仰、横摇）；旋转=旋转矩阵（偏航、俯仰、横摇）；%构造一个摄像机矩阵聚焦长度=[309.4362344.2161]；主点=[318.9034257.5352]；歪斜=0；camMatrix=[旋转；平移]*内部矩阵（聚焦长度，…倾斜，主点）；%将camMatrix转换为二维单应性tform2D=[camMatrix（1，：）；camMatrix（2，：）；camMatrix（4，：）]；%下降Z t形式=投影2D（t或M2D）；tform=tform.invert（）；端部功能平移=平移向量（偏航、俯仰、滚转）传感器位置=[0]；高度=2.1798；%离地面旋转矩阵的安装高度（米）=（…旋转（偏航）*…%最后一次旋转旋转（90节）*…旋转（滚转）…%第一次旋转）；%通过添加平移sl=传感器位置来调整传感器位置；转换单位=[sl（2），sl（1），高度]；平移=平移世界单位*旋转矩阵；结束%-------------------------------------------------------------------------------------绕X轴旋转%函数R=rotX（a）a=deg2rad（a）；R=[…10 0；0 cos（a）-sin（a）；0 sin（a）cos（a）]；结束%-------------------------------------------------------------------------------------绕Y轴旋转%函数R=rotY（a）a=deg2rad（a）；R=[…cos（a）0sin（a）；0110；-sin（a）0cos（a）]；结束%-------------------------------------------------------------------------------------绕Z轴旋转%函数R=rotZ（a）a=deg2rad（a）；R=[…cos（a）-sin（a）0；sin（a）cos（a）0；01]；end%-------------------------------------------------------------------------------------%给定偏航、俯仰和横摇，确定适当的欧拉%角度以及它们应用于%将摄像头坐标系与车辆坐标系%对齐的顺序。生成的矩阵是一个旋转矩阵，它与平移向量一起定义相机的外部参数。函数旋转=旋转矩阵（偏航、俯仰、滚转）旋转=（…旋转（180）*…%上次旋转：点Z向上旋转（-90）*…%X-Y交换旋转（偏航）*…%点相机向前旋转（90俯仰）*…%取消俯仰旋转（滚转）.%第一次旋转：“取消滚转”）；结束函数intrinsicMat=intrinsicMatrix（焦点长度、偏移、主点）intrinsicMat=。。。[FocalLength（1），0，0；…歪斜，FocalLength（2），0；…主点（1），主点（2），1]；终止

代码生成成功:要查看报告，打开('codegen/lib/detect_lane/html/report.mldatx')。

．cnn_lanenet_data_offset.bin . .cnn_lanenet_data_scale.bin DeepLearningNetwork。cnn_lanenet_fc6_b.bin DeepLearningNetwork.h cnn_lanenet_fc6_w.bin DeepLearningNetwork.ho cnn_lanenet_fcLane1_b.bin MWCNNLayerImpl。cnn_lanenet_fclone1_w .bin MWCNNLayerImpl.hpp cnn_lanenet_fclone2_b .bino cnn_lanenet_fcLane2_w.bin MWCudaDimUtility。cu cnn_lanenet_responsename .txt mwcudadimultiple .hpp codeInfo. txt。垫MWCudaDimUtility。o codedescriptor。dmr mwelementwiseaffinlayer .cpp coder_array.h mwelementwiseaffinlayer .hpp compileInfo。垫MWElementwiseAffineLayer。o detect_lane。MWElementwiseAffineLayerImpl。铜detect_lane。hpp detect_lane.h MWElementwiseAffineLayerImpl. cu MWElementwiseAffineLayerImpl. cuo detect_lane。o MWElementwiseAffineLayerImplKernel。铜detect_lane_data。cu MWElementwiseAffineLayerImplKernel.o detect_lane_data.h MWFusedConvReLULayer.cpp detect_lane_data.o MWFusedConvReLULayer.hpp detect_lane_initialize.cu MWFusedConvReLULayer.o detect_lane_initialize.h MWFusedConvReLULayerImpl.cu detect_lane_initialize.o MWFusedConvReLULayerImpl.hpp detect_lane_ref.rsp MWFusedConvReLULayerImpl.o detect_lane_rtw.mk MWKernelHeaders.hpp detect_lane_rtwutil.cu MWTargetNetworkImpl.cu detect_lane_rtwutil.h MWTargetNetworkImpl.hpp detect_lane_rtwutil.o MWTargetNetworkImpl.o detect_lane_terminate.cu buildInfo.mat detect_lane_terminate.h cnn_api.cpp detect_lane_terminate.o cnn_api.hpp detect_lane_types.h cnn_api.o examples cnn_lanenet_conv1_b.bin gpu_codegen_info.mat cnn_lanenet_conv1_w.bin html cnn_lanenet_conv2_b.bin interface cnn_lanenet_conv2_w.bin predict.cu cnn_lanenet_conv3_b.bin predict.h cnn_lanenet_conv3_w.bin predict.o cnn_lanenet_conv4_b.bin rtw_proj.tmw cnn_lanenet_conv4_w.bin rtwtypes.h cnn_lanenet_conv5_b.bin cnn_lanenet_conv5_w.bin

/* The MathWorks, Inc. */ #include  #include  #include  #include "opencv2/opencv.hpp" #include  #include  #include "detect_lane.h" using namespace cv;void readData(float *input, mat&orig, Mat& im) {Size Size(227,227); / /输入数据调整(源自,im,大小,0,0,INTER_LINEAR);(int j = 0; < 227 * 227; j + +) {/ / BGR RGB输入[2 * 227 * 227 + j] =(浮动)(im.data [j * 3 + 0]);输入(1 * 227 * 227 + j] =(浮动)(im.data [j * 3 + 1]);输入[0 * 227 * 227 + j] =(浮动)(im.data [j * 3 + 2]);}} void addLane(float pts[28][2]， Mat & im, int numPts) {std::vector iArray;for (int k = 0;k < numPts;k + +) {iArray.push_back (Point2f (pts [k] [0], pts [k] [1])); } Mat curve(iArray, true); curve.convertTo(curve, CV_32S); //adapt type for polylines polylines(im, curve, false, CV_RGB(255,255,0), 2, CV_AA); } void writeData(float *outputBuffer, Mat & im, int N, double means[6], double stds[6]) { // get lane coordinates boolean_T laneFound = 0; float ltPts[56]; float rtPts[56]; detect_lane(outputBuffer, means, stds, &laneFound, ltPts, rtPts); if (!laneFound) { return; } float ltPtsM[28][2]; float rtPtsM[28][2]; for(int k=0; k<28; k++) { ltPtsM[k][0] = ltPts[k]; ltPtsM[k][1] = ltPts[k+28]; rtPtsM[k][0] = rtPts[k]; rtPtsM[k][1] = rtPts[k+28]; } addLane(ltPtsM, im, 28); addLane(rtPtsM, im, 28); } void readMeanAndStds(const char* filename, double means[6], double stds[6]) { FILE* pFile = fopen(filename, "rb"); if (pFile==NULL) { fputs ("File error",stderr); return; } // obtain file size fseek (pFile , 0 , SEEK_END); long lSize = ftell(pFile); rewind(pFile); double* buffer = (double*)malloc(lSize); size_t result = fread(buffer,sizeof(double),lSize,pFile); if (result*sizeof(double) != lSize) { fputs ("Reading error",stderr); return; } for (int k = 0 ; k < 6; k++) { means[k] = buffer[k]; stds[k] = buffer[k+6]; } free(buffer); } // Main function int main(int argc, char* argv[]) { float *inputBuffer = (float*)calloc(sizeof(float),227*227*3); float *outputBuffer = (float*)calloc(sizeof(float),6); if ((inputBuffer == NULL) || (outputBuffer == NULL)) { printf("ERROR: Input/Output buffers could not be allocated!\n"); exit(-1); } // get ground truth mean and std double means[6]; double stds[6]; readMeanAndStds("mean.bin", means, stds); if (argc < 2) { printf("Pass in input video file name as argument\n"); return -1; } VideoCapture cap(argv[1]); if (!cap.isOpened()) { printf("Could not open the video capture device.\n"); return -1; } cudaEvent_t start, stop; float fps = 0; cudaEventCreate(&start); cudaEventCreate(&stop); Mat orig, im; namedWindow("Lane detection demo",CV_WINDOW_NORMAL); while(true) { cudaEventRecord(start); cap >> orig; if (orig.empty()) break; readData(inputBuffer, orig, im); writeData(inputBuffer, orig, 6, means, stds); cudaEventRecord(stop); cudaEventSynchronize(stop); char strbuf[50]; float milliseconds = -1.0; cudaEventElapsedTime(&milliseconds, start, stop); fps = fps*.9+1000.0/milliseconds*.1; sprintf (strbuf, "%.2f FPS", fps); putText(orig, strbuf, cvPoint(200,30), CV_FONT_HERSHEY_DUPLEX, 1, CV_RGB(0,0,0), 2); imshow("Lane detection demo", orig); if( waitKey(50)%256 == 27 ) break; // stop capturing by pressing ESC */ } destroyWindow("Lane detection demo"); free(inputBuffer); free(outputBuffer); return 0; }