目标检测 YOLOv5 – 数据增强

flyfish

版本 YOLOv5:v5
原始代码 YOLOv5
源码镜像（可以快速打开）
本文所
描述的数据增强包含一下12个超参数，参数存储在hyp.scratch.yaml文件中

# hsv_h: 0.015  # image HSV-Hue augmentation (fraction)
# hsv_s: 0.7  # image HSV-Saturation augmentation (fraction)
# hsv_v: 0.4  # image HSV-Value augmentation (fraction)
# degrees: 0.0  # image rotation (+/- deg)
# translate: 0.1  # image translation (+/- fraction)
# scale: 0.5  # image scale (+/- gain)
# shear: 0.0  # image shear (+/- deg)
# perspective: 0.0  # image perspective (+/- fraction), range 0-0.001   
# flipud: 0.0  # image flip up-down (probability)
# fliplr: 0.5  # image flip left-right (probability) 
# mosaic: 1.0  # image mosaic (probability)
# mixup: 0.0  # image mixup (probability)

一 Hue、Saturation、Value

# hsv_h: 0.015  # image HSV-Hue augmentation (fraction)
# hsv_s: 0.7  # image HSV-Saturation augmentation (fraction)
# hsv_v: 0.4  # image HSV-Value augmentation (fraction)

代码实现

import numpy as np
import cv2
def augment_hsv(img, hgain=0.015, sgain=0.7, vgain=0.4):
    r = np.random.uniform(-1, 1, 3) * [hgain, sgain, vgain] + 1  # random gains
    hue, sat, val = cv2.split(cv2.cvtColor(img, cv2.COLOR_BGR2HSV))
    dtype = img.dtype  # uint8

    x = np.arange(0, 256, dtype=np.int16)
    lut_hue = ((x * r[0]) % 180).astype(dtype)
    lut_sat = np.clip(x * r[1], 0, 255).astype(dtype)
    lut_val = np.clip(x * r[2], 0, 255).astype(dtype)

    img_hsv = cv2.merge((cv2.LUT(hue, lut_hue), cv2.LUT(sat, lut_sat), cv2.LUT(val, lut_val))).astype(dtype)
    cv2.cvtColor(img_hsv, cv2.COLOR_HSV2BGR, dst=img)  # no return needed

path="/coco128/images/train2017/000000000641.jpg"
original_img = cv2.imread(path)
cv2.imshow("original:",original_img)
cv2.waitKey(0)
augment_hsv(original_img)
cv2.imshow("augment_hsv:",original_img)
cv2.waitKey(0)

原图

增强结果

二 degrees、translate、scale、perspective

# degrees: 0.0  # image rotation (+/- deg)
# translate: 0.1  # image translation (+/- fraction)
# scale: 0.5  # image scale (+/- gain)
# shear: 0.0  # image shear (+/- deg)
# perspective: 0.0  # image perspective (+/- fraction), range 0-0.001   

源码实现

import math
import random
import numpy as np
import cv2

def random_perspective(img, degrees=10, translate=.1, scale=.1, shear=10, perspective=0.0,
                       border=(0, 0)):

    height = img.shape[0] + border[0] * 2  # shape(h,w,c)
    width = img.shape[1] + border[1] * 2

    # Center
    C = np.eye(3)
    C[0, 2] = -img.shape[1] / 2  # x translation (pixels)
    C[1, 2] = -img.shape[0] / 2  # y translation (pixels)

    # Perspective
    P = np.eye(3)
    P[2, 0] = random.uniform(-perspective, perspective)  # x perspective (about y)
    P[2, 1] = random.uniform(-perspective, perspective)  # y perspective (about x)

    # Rotation and Scale
    R = np.eye(3)
    a = random.uniform(-degrees, degrees)
    # a += random.choice([-180, -90, 0, 90])  # add 90deg rotations to small rotations
    s = random.uniform(1 - scale, 1 + scale)
    # s = 2 ** random.uniform(-scale, scale)
    R[:2] = cv2.getRotationMatrix2D(angle=a, center=(0, 0), scale=s)

    # Shear
    S = np.eye(3)
    S[0, 1] = math.tan(random.uniform(-shear, shear) * math.pi / 180)  # x shear (deg)
    S[1, 0] = math.tan(random.uniform(-shear, shear) * math.pi / 180)  # y shear (deg)

    # Translation
    T = np.eye(3)
    T[0, 2] = random.uniform(0.5 - translate, 0.5 + translate) * width  # x translation (pixels)
    T[1, 2] = random.uniform(0.5 - translate, 0.5 + translate) * height  # y translation (pixels)

    # Combined rotation matrix
    M = T @ S @ R @ P @ C  # order of operations (right to left) is IMPORTANT
    if (border[0] != 0) or (border[1] != 0) or (M != np.eye(3)).any():  # image changed
        if perspective:
            img = cv2.warpPerspective(img, M, dsize=(width, height), borderValue=(114, 114, 114))
        else:  # affine
            img = cv2.warpAffine(img, M[:2], dsize=(width, height), borderValue=(114, 114, 114))

    return img


path="/coco128/images/train2017/000000000641.jpg"
original_img = cv2.imread(path)
img=random_perspective(original_img, degrees=50, translate=0., scale=.0, shear=0, perspective=0,
                       border=(0, 0))
cv2.imshow("random_perspective_degrees:",img)
cv2.waitKey(0)

img=random_perspective(original_img, degrees=0, translate=0.5, scale=.0, shear=0, perspective=0,
                       border=(0, 0))
cv2.imshow("random_perspective_translate:",img)
cv2.waitKey(0)

img=random_perspective(original_img, degrees=0, translate=0.0, scale=0.9, shear=0, perspective=0,
                       border=(0, 0))
cv2.imshow("random_perspective_scale:",img)
cv2.waitKey(0)

img=random_perspective(original_img, degrees=0, translate=0.0, scale=0, shear=20, perspective=0,
                       border=(0, 0))
cv2.imshow("random_perspective_shear:",img)
cv2.waitKey(0)

img=random_perspective(original_img, degrees=0, translate=0.0, scale=0, shear=0, perspective=0.001,
                       border=(0, 0))
cv2.imshow("random_perspective_perspective:",img)
cv2.waitKey(0)

增强结果

shear

degrees

translate

关于图像translate平移的实现
实现方式1

# 使用imutils.translate函数
import argparse
import cv2
import imutils
import numpy as np

ap = argparse.ArgumentParser()
ap.add_argument("--image", type=str, default="1.jpg",
                help="input image")
args = vars(ap.parse_args())
image = cv2.imread(args["image"])
cv2.imshow("Original", image)

# 正数是向右，向下移动
# 负数是向左，向上移动
shifted = imutils.translate(image, 200, 100)
cv2.imshow("Right Down", shifted)
cv2.waitKey(0)
cv2.destroyAllWindows()

实现方式2

#使用cv2.warpAffine实现
#向右移动100像素,像下移动200像素
M = np.float32([[1, 0, 100], [0, 1, 200]])
shifted = cv2.warpAffine(image, M, (image.shape[1], image.shape[0]))
cv2.imshow("Right Down", shifted)

# 向左移动100像素，向上移动200像素
M = np.float32([[1, 0, -100], [0, 1, -200]])
shifted = cv2.warpAffine(image, M, (image.shape[1], image.shape[0]))
cv2.imshow("Up Left", shifted)

scale

perspective

三 flip up-down、flip left-right

代码实现

import numpy as np
import cv2

path="/coco128/images/train2017/000000000641.jpg"
original_img = cv2.imread(path)
img = np.flipud(original_img)
cv2.imshow("flip up-down:",img)
cv2.waitKey(0)

img = np.fliplr(original_img)
cv2.imshow("flip left-right:",img)
cv2.waitKey(0)

增强结果

flip up-down

flip left-right

四 mosaic

四张图像拼成一张图像
下面的16张图像是一个batch-size
实例1

实例2

五 mixup

在YOLOv5中只有启用了mosaic才有机会启用mixup
MixUp


想要实现数据增强，只需要更改配置文件即可完成
例如想要图片有50%的概率上下翻转和左右翻转
只需要在hyp.scratch.yaml文件中，设置flipud和fliplr的数值即可

flipud: 0.5  # image flip up-down (probability)
fliplr: 0.5  # image flip left-right (probability) 

阅读此文章的人还阅读了以下内容
目标检测 YOLOv5 – 卷积层和BN层的融合
目标检测 YOLOv5 – Sample Assignment
目标检测 YOLOv5 – 数据增强
目标检测 YOLOv5 – 学习率
目标检测 YOLOv5 – 多机多卡训练
目标检测 YOLOv5 – 浮点取模
目标检测 YOLOv5 – 在多类别中应用NMS（非极大值抑制）
目标检测 YOLOv5 – loss for objectness and classification
目标检测 YOLOv5 – loss for bounding box regression
目标检测 YOLOv5 – 指标计算
目标检测 YOLOv5 – anchor设置
目标检测 YOLOv5 – SPP模块
目标检测 YOLOv5 – 边框预测（bounding box prediction）
目标检测 YOLOv5 – 自定义网络结构（YOLOv5-ShuffleNetV2）
目标检测 YOLOv5 – 常见的边框（bounding box ）坐标表示方法
目标检测 YOLOv5 – 图像大小与loss权重的关系
目标检测 YOLOv5 – 根据配置改变网络的深度和宽度
目标检测 YOLOv5 – 转ncnn移动端部署
目标检测 YOLOv5 – Backbone中的Focus
目标检测 YOLOv5 – 模型训练、推理、导出命令
目标检测 YOLOv5 – 人脸数据集widerface转YOLOv5格式
目标检测 YOLOv5 – 使用的数据集格式
目标检测 YOLOv5 – 中使用COCO数据集
目标检测 YOLOv5 – CrowdHuman数据集格式转YOLOv5格式

来源：西西弗Sisyphus