yolo 学习系列（五）：K-means维度聚类

yolo 学习系列（五）：K-means维度聚类

《深度学习：手把手教你做目标检测（YOLO、SSD）》视频教程

1. 维度聚类

1.1 聚类目的

在Ubuntu系统下运行voc_label.py生成训练集和测试集列表文件，在Windows下生成的会有编码错误

1. 当使用自己的数据集进行 YOLO 训练时，首先要做的就是讲 anchors 更换为自己的宽高尺寸，默认的是 VOC/COCO 数据集中 20/80 类目标的宽高尺寸。
   维度聚类就是利用训练集迭代找出该数据集的候选框宽高尺寸。
2. 先放结果

*****************
n_anchors = 5
*****************

k-means result：

(9.503125000000356, 8.046875000000572)
k-means result：

(4.367264851484575, 6.158106435643181)
k-means result：

(7.935546874999882, 4.897786458333437)
k-means result：

(6.179457720588269, 7.807674632352704)
k-means result：

(3.7825989208630717, 3.5274280575536627)

*****************
n_anchors = 6
*****************

k-means result：

(6.195027372262894, 7.749087591240633)
k-means result：

(9.131578947368867, 9.546875000000345)
k-means result：

(4.324420103092257, 6.20972938144295)
k-means result：

(9.331640624999876, 5.898828124999875)
k-means result：

(3.7665307971012534, 3.5264945652170954)
k-means result：

(7.337187499999761, 4.57375000000048)

*****************
n_anchors = 9
*****************

k-means result：

(2.6570723684221056, 3.3848684210523685)
k-means result：

(3.6341911764701336, 5.45174632352828)
k-means result：

(5.790958737863881, 8.205703883494753)
k-means result：

(7.290719696969482, 4.354640151515531)
k-means result：

(5.568824404761428, 5.959821428571835)
k-means result：

(9.106445312500655, 9.826171875000409)
k-means result：

(9.356770833333087, 5.37803819444432)
k-means result：

(8.429036458334373, 7.22656250000046)
k-means result：

(4.455696202530808, 3.311313291139051)

1.2 源码

源码-Github

# coding=utf-8
# k-means ++ for YOLOv2 anchors
# 通过k-means ++ 算法获取YOLOv2需要的anchors的尺寸
import numpy as np

# 定义Box类，描述bounding box的坐标
class Box():
    def __init__(self, x, y, w, h):
        self.x = x
        self.y = y
        self.w = w
        self.h = h


# 计算两个box在某个轴上的重叠部分
# x1是box1的中心在该轴上的坐标
# len1是box1在该轴上的长度
# x2是box2的中心在该轴上的坐标
# len2是box2在该轴上的长度
# 返回值是该轴上重叠的长度
def overlap(x1, len1, x2, len2):
    len1_half = len1 / 2
    len2_half = len2 / 2

    left = max(x1 - len1_half, x2 - len2_half)
    right = min(x1 + len1_half, x2 + len2_half)

    return right - left


# 计算box a 和box b 的交集面积
# a和b都是Box类型实例
# 返回值area是box a 和box b 的交集面积
def box_intersection(a, b):
    w = overlap(a.x, a.w, b.x, b.w)
    h = overlap(a.y, a.h, b.y, b.h)
    if w < 0 or h < 0:
        return 0

    area = w * h
    return area


# 计算 box a 和 box b 的并集面积
# a和b都是Box类型实例
# 返回值u是box a 和box b 的并集面积
def box_union(a, b):
    i = box_intersection(a, b)
    u = a.w * a.h + b.w * b.h - i
    return u


# 计算 box a 和 box b 的 iou
# a和b都是Box类型实例
# 返回值是box a 和box b 的iou
def box_iou(a, b):
    return box_intersection(a, b) / box_union(a, b)


# 使用k-means ++ 初始化 centroids，减少随机初始化的centroids对最终结果的影响
# boxes是所有bounding boxes的Box对象列表
# n_anchors是k-means的k值
# 返回值centroids 是初始化的n_anchors个centroid
def init_centroids(boxes,n_anchors):
    centroids = []
    boxes_num = len(boxes)

    centroid_index = np.random.choice(boxes_num, 1)
    centroids.append(boxes[centroid_index])

    print(centroids[0].w,centroids[0].h)

    for centroid_index in range(0,n_anchors-1):

        sum_distance = 0
        distance_thresh = 0
        distance_list = []
        cur_sum = 0

        for box in boxes:
            min_distance = 1
            for centroid_i, centroid in enumerate(centroids):
                distance = (1 - box_iou(box, centroid))
                if distance < min_distance:
                    min_distance = distance
            sum_distance += min_distance
            distance_list.append(min_distance)

        distance_thresh = sum_distance*np.random.random()

        for i in range(0,boxes_num):
            cur_sum += distance_list[i]
            if cur_sum > distance_thresh:
                centroids.append(boxes[i])
                print(boxes[i].w, boxes[i].h)
                break

    return centroids


# 进行 k-means 计算新的centroids
# boxes是所有bounding boxes的Box对象列表
# n_anchors是k-means的k值
# centroids是所有簇的中心
# 返回值new_centroids 是计算出的新簇中心
# 返回值groups是n_anchors个簇包含的boxes的列表
# 返回值loss是所有box距离所属的最近的centroid的距离的和
def do_kmeans(n_anchors, boxes, centroids):
    loss = 0
    groups = []
    new_centroids = []
    for i in range(n_anchors):
        groups.append([])
        new_centroids.append(Box(0, 0, 0, 0))

    for box in boxes:
        min_distance = 1
        group_index = 0
        for centroid_index, centroid in enumerate(centroids):
            distance = (1 - box_iou(box, centroid))
            if distance < min_distance:
                min_distance = distance
                group_index = centroid_index
        groups[group_index].append(box)
        loss += min_distance
        new_centroids[group_index].w += box.w
        new_centroids[group_index].h += box.h

    for i in range(n_anchors):
        new_centroids[i].w /= len(groups[i])
        new_centroids[i].h /= len(groups[i])

    return new_centroids, groups, loss


# 计算给定bounding boxes的n_anchors数量的centroids
# label_path是训练集列表文件地址
# n_anchors 是anchors的数量
# loss_convergence是允许的loss的最小变化值
# grid_size * grid_size 是栅格数量
# iterations_num是最大迭代次数
# plus = 1时启用k means ++ 初始化centroids
def compute_centroids(label_path,n_anchors,loss_convergence,grid_size,iterations_num,plus):

    boxes = []
    label_files = []
    f = open(label_path)
    for line in f:
        label_path = line.rstrip().replace('images', 'labels')
        label_path = label_path.replace('JPEGImages', 'labels')
        label_path = label_path.replace('.jpg', '.txt')
        label_path = label_path.replace('.JPEG', '.txt')
        label_files.append(label_path)
    f.close()

    for label_file in label_files:
        f = open(label_file)
        for line in f:
            temp = line.strip().split(" ")
            if len(temp) > 1:
                boxes.append(Box(0, 0, float(temp[3]), float(temp[4])))

    if plus:
        centroids = init_centroids(boxes, n_anchors)
    else:
        centroid_indices = np.random.choice(len(boxes), n_anchors)
        centroids = []
        for centroid_index in centroid_indices:
            centroids.append(boxes[centroid_index])

    # iterate k-means
    centroids, groups, old_loss = do_kmeans(n_anchors, boxes, centroids)
    iterations = 1
    while (True):
        centroids, groups, loss = do_kmeans(n_anchors, boxes, centroids)
        iterations = iterations + 1
        print("loss = %f" % loss)
        if abs(old_loss - loss) < loss_convergence or iterations > iterations_num:
            break
        old_loss = loss

        for centroid in centroids:
            print(centroid.w * grid_size, centroid.h * grid_size)

    # print result
    for centroid in centroids:
        print("k-means result：\n")
        print(centroid.w * grid_size, centroid.h * grid_size)


label_path = "/home/chris/workspace/2007_train.txt"
n_anchors = 5
loss_convergence = 1e-6
grid_size = 13
iterations_num = 100
plus = 0
compute_centroids(label_path,n_anchors,loss_convergence,grid_size,iterations_num,plus)

1.3 运行结果

k-means result：

(8.979910714285644, 5.140624999999976)
k-means result：

(4.5747282608690005, 7.813858695652043)
k-means result：

(2.2546296296290005, 7.7939814814810005)
k-means result：

(11.235351562499998, 9.699218750000407)
k-means result：

(2.442095588236353, 3.5698529411762943)

1.4 参考资源

YOLOv2通过k-means来获取anchor boxes

2.Caffe-SSD长宽比聚类

k-means 计算voc2012数据集的检测anchors的长宽聚类结果

本文用到了两个程序，一个是 get_w_h.py 用于获取数据集目标的宽和高（原因是参考程序该部分功能有问题），第二个是参考链接里的聚类绘图程序。

通过 get_w_h.py 分别获取目标的宽和高，并通过 excel 制作成第一列为宽，第二列为高的 data1.txt 文件，并将其与 K-means.py 文件存放在同一目录下，运行该 py 文件即可。

#  第67行
#      print(ob_w)
      print(ob_h)