YOLO源码详解(五)-追本溯源7*7个grid

来源:互联网 发布:mac qq能远程协助吗 编辑:程序博客网 时间:2024/06/06 01:26

作者:木凌
时间:2016年11月。
文章连接:http://blog.csdn.net/u014540717


微信群
这里写图片描述


最近一直有人在问,把图像分为7*7个网格,每个网格推荐两个框是什么意思,一直没搞明白,今天我们就从源码入手,追本溯源,彻底理解7*7个grid
在YOLO源码详解(三)- 前向传播(forward)里,我们分析了detection_layer.c层的代码,我们来看一下truth_index是怎么定义的:

int truth_index = (b*locations + i)*(1+l.coords+l.classes);

这里参数意义如下:
locations:7*7
b :batch size的索引
i :locations的索引
1 :置信度
l.coords :值为4,分别表示x,y,w,h
l.classes : 20

然后在下面我们可以看到如下代码段

//l.n就是一个网格要推荐几个框,论文中l.n=2for(j = 0; j < l.n; ++j){                    int box_index = index + locations*(l.classes + l.n) + (i*l.n + j) * l.coords;                    box out = float_to_box(l.output + box_index);                    out.x /= l.side;                    out.y /= l.side;                    if (l.sqrt){                        out.w = out.w*out.w;                        out.h = out.h*out.h;                    }                    //计算iou的值                    float iou  = box_iou(out, truth);                    //iou = 0;                    //计算均方根误差(root-mean-square error)                    float rmse = box_rmse(out, truth);                    //选出iou最大或者均方根误差最小的那个框作为最后预测框~                    if(best_iou > 0 || iou > 0){                        if(iou > best_iou){                            best_iou = iou;                            best_index = j;                        }                    }else{                        if(rmse < best_rmse){                            best_rmse = rmse;                            best_index = j;                        }                    }                }

上述代码中最重要的是box_iou(out, truth);这句代码,这句代码是要计算你输出的框和真实框的IOU,truth的定义如下:

box truth = float_to_box(state.truth + truth_index + 1 + l.classes);

从定义我们可以得到,真实框的坐标来自state.truth,我们来追本溯源
state.truth最初是在network.c中赋值的

//network.cstate.truth = y;

y又从哪里来呢?最初的y是从load_data_in_thread(args);这个函数中获得的,我们来剖析下该函数

//data.cpthread_t load_data_in_thread(load_args args){    pthread_t thread;    struct load_args *ptr = calloc(1, sizeof(struct load_args));    *ptr = args;    //调用load_thread这个函数    if(pthread_create(&thread, 0, load_thread, ptr)) error("Thread creation failed");    return thread;}
//data.cvoid *load_thread(void *ptr){    //printf("Loading data: %d\n", rand());    load_args a = *(struct load_args*)ptr;    if(a.exposure == 0) a.exposure = 1;    if(a.saturation == 0) a.saturation = 1;    if(a.aspect == 0) a.aspect = 1;    if (a.type == OLD_CLASSIFICATION_DATA){        *a.d = load_data_old(a.paths, a.n, a.m, a.labels, a.classes, a.w, a.h);    } else if (a.type == CLASSIFICATION_DATA){        *a.d = load_data_augment(a.paths, a.n, a.m, a.labels, a.classes, a.hierarchy, a.min, a.max, a.size, a.angle, a.aspect, a.hue, a.saturation, a.exposure);    } else if (a.type == SUPER_DATA){        *a.d = load_data_super(a.paths, a.n, a.m, a.w, a.h, a.scale);    } else if (a.type == WRITING_DATA){        *a.d = load_data_writing(a.paths, a.n, a.m, a.w, a.h, a.out_w, a.out_h);    } else if (a.type == REGION_DATA){        //因为a.type == REGION_DATA,所以调用这个函数,我们继续追~        *a.d = load_data_region(a.n, a.paths, a.m, a.w, a.h, a.num_boxes, a.classes, a.jitter, a.hue, a.saturation, a.exposure);...
//data.cdata load_data_region(int n, char **paths, int m, int w, int h, int size, int classes, float jitter, float hue, float saturation, float exposure){    char **random_paths = get_random_paths(paths, n, m);    int i;    data d = {0};    d.shallow = 0;    //n就是batch size啦    d.X.rows = n;    //给X(也就是图像数据)分配内存    d.X.vals = calloc(d.X.rows, sizeof(float*));    d.X.cols = h*w*3;    int k = size*size*(5+classes);    //终于找到你啦~\(≧▽≦)/~。这里先给y分配了内存,注意一共分配了n*k个float类型的内存块,为什么分配这么多呢?慢慢往下看~    d.y = make_matrix(n, k);    for(i = 0; i < n; ++i){        //读取图像        image orig = load_image_color(random_paths[i], 0, 0);        int oh = orig.h;        int ow = orig.w;        //这里jitter=0.2(cfg文件中有写),这就是所谓的抖动了,其实就是crop(数据增广的一种)        //剪掉的不能太多,这里设置图像的左边和右边最多剪掉dw(整幅图像宽度的1/5),上边和下边最多剪掉dh(整幅图像高度的1/5)        int dw = (ow*jitter);        int dh = (oh*jitter);        //rand_uniform生成(-dw, dw)的一个随机数        int pleft  = rand_uniform(-dw, dw);        int pright = rand_uniform(-dw, dw);        int ptop   = rand_uniform(-dh, dh);        int pbot   = rand_uniform(-dh, dh);        //swidth是图像剪完后的宽度,sheight是图像剪完后的高度        int swidth =  ow - pleft - pright;        int sheight = oh - ptop - pbot;        //sx是图像剪完后宽度和原始图像的宽度比,同理sy        float sx = (float)swidth  / ow;        float sy = (float)sheight / oh;        //设置图像随机翻转        int flip = rand()%2;        //开始剪切图像,咔咔咔,具体代码不看了,很简单~        image cropped = crop_image(orig, pleft, ptop, swidth, sheight);        //dx=pleft/swidth,dy=ptop/sheight        float dx = ((float)pleft/ow)/sx;        float dy = ((float)ptop /oh)/sy;        //都剪完了,当然要把图像重新resize到448*448(论文中说了,输入图像是448*448)        image sized = resize_image(cropped, w, h);        //翻转图像~        if(flip) flip_image(sized);        //图像随机排序        random_distort_image(sized, hue, saturation, exposure);        //最终d.X.vals[]存储的就是要输入的数据啦,准备好X了,我们去准备下y        d.X.vals[i] = sized.data;        //开始追y,追追追~        fill_truth_region(random_paths[i], d.y.vals[i], classes, size, flip, dx, dy, 1./sx, 1./sy);...
//data.cvoid fill_truth_region(char *path, float *truth, int classes, int num_boxes, int flip, float dx, float dy, float sx, float sy){    char labelpath[4096];    //有人一直不知道labels怎么来的,说源码都没设置labels的路径啊,怎么读的labels啊,那不是成了无监督学习?其实源码只是没直接设置labels的路径而已,把images替换为labels,在把.jpg替换为.txt,labels的路径就有了~    find_replace(path, "images", "labels", labelpath);    find_replace(labelpath, "JPEGImages", "labels", labelpath);    find_replace(labelpath, ".jpg", ".txt", labelpath);    find_replace(labelpath, ".png", ".txt", labelpath);    find_replace(labelpath, ".JPG", ".txt", labelpath);    find_replace(labelpath, ".JPEG", ".txt", labelpath);    int count = 0;    //从.txt中读取labels值,count记录框的个数    box_label *boxes = read_boxes(labelpath, &count);    //把框随机排序~    randomize_boxes(boxes, count);    //因为图像已经被修剪了,所以框的坐标也要改一改,correct_boxes函数就是把框在原始图像下的坐标转到修剪后图像下的坐标    correct_boxes(boxes, count, dx, dy, sx, sy, flip);    float x,y,w,h;    int id;    int i;    for (i = 0; i < count; ++i) {        x =  boxes[i].x;        y =  boxes[i].y;        w =  boxes[i].w;        h =  boxes[i].h;        id = boxes[i].id;        //修剪后,太小的框就不作为正样本了        if (w < .01 || h < .01) continue;        //这里x的值为0~1之间(不一定能取到0和1,因为图像被修剪过了,坐标的范围也变了),num_boxes=7,所以col和row都是0~6之间的整数        int col = (int)(x*num_boxes);        int row = (int)(y*num_boxes);        //x和y又被打回原型,又变成0~1之间的数了        x = x*num_boxes - col;        y = y*num_boxes - row;        //index就懂了吧,一共7*7个网格,每个网格的索引是0~6        int index = (col+row*num_boxes)*(5+classes);        if (truth[index]) continue;        //如果第i个框落在这个网格里,就把相应的置信度赋1        truth[index++] = 1;        //然后看标签id是几,就把对应的类别处赋值为1        if (id < classes) truth[index+id] = 1;        index += classes;        //再赋值框的x,y,w,h到truth        truth[index++] = x;        truth[index++] = y;        truth[index++] = w;        truth[index++] = h;    }    free(boxes);}

y值追完了,我们再回过头来看

float iou  = box_iou(out, truth);

out(每个网格一共l.n个out,论文中l.n=2)就是网络回归出来的值,然后把out的值和truth中的值对应比较,计算出iou,然后从l.n个iou中挑出iou最高的一个,作为最后的预测框,说白了就是:只有该框会对loss function产生影响,其他框不产生影响,仅此而已。

现在你知道7*7个网格,每个网格推荐两个框是怎么回事儿了吗?


(END)

1 0
原创粉丝点击