標題HBB格式的數據擴增
該格式下目標被標註爲xmin，ymin，xmax，ymax
import os
import cv2
import xml.dom.minidom
from xml.dom.minidom import Document
import math
#獲取路徑下所有文件的完整路徑，用於讀取文件用
def GetFileFromThisRootDir(dir,ext = None):
allfiles = []
needExtFilter = (ext != None)
for root,dirs,files in os.walk(dir):
for filespath in files:
filepath = os.path.join(root, filespath)
extension = os.path.splitext(filepath)[1][1:]
if needExtFilter and extension in ext:
allfiles.append(filepath)
elif not needExtFilter:
allfiles.append(filepath)
return allfiles
def limit_value(a,b):
if a<1:
a = 1
if a>b:
a = b-1
return a
#讀取xml文件，xmlfile參數表示xml的路徑
def readXml(xmlfile):
DomTree = xml.dom.minidom.parse(xmlfile)
annotation = DomTree.documentElement
sizelist = annotation.getElementsByTagName('size') #[<DOM Element: filename at 0x381f788>]
heights = sizelist[0].getElementsByTagName('height')
height = int(heights[0].childNodes[0].data)
widths =sizelist[0].getElementsByTagName('width')
width = int(widths[0].childNodes[0].data)
depths = sizelist[0].getElementsByTagName('depth')
depth = int(depths[0].childNodes[0].data)
objectlist = annotation.getElementsByTagName('object')
bboxes = []
for objects in objectlist:
namelist = objects.getElementsByTagName('name')
class_label = namelist[0].childNodes[0].data
bndbox = objects.getElementsByTagName('bndbox')[0]
x1_list = bndbox.getElementsByTagName('xmin')
x1 = int(float(x1_list[0].childNodes[0].data))
y1_list = bndbox.getElementsByTagName('ymin')
y1 = int(float(y1_list[0].childNodes[0].data))
x2_list = bndbox.getElementsByTagName('xmax')
x2 = int(float(x2_list[0].childNodes[0].data))
y2_list = bndbox.getElementsByTagName('ymax')
y2 = int(float(y2_list[0].childNodes[0].data))
#這裏我box的格式【xmin，ymin，xmax，ymax，classname】
bbox = [x1,y1,x2,y2,class_label]
bboxes.append(bbox)
return bboxes,width,height,depth
#圖像旋轉用，裏面的angle是角度制的
def im_rotate(im,angle,center = None,scale = 1.0):
h,w = im.shape[:2]
if center is None:
center = (w/2,h/2)
M = cv2.getRotationMatrix2D(center,angle,scale)
im_rot = cv2.warpAffine(im,M,(w,h))
return im_rot
def im_flip(im, method = 'H'): #翻轉圖像
if method == 'H': # Flipped Horizontally 水平翻轉
im_flip = cv2.flip(im, 1)
elif method == 'V': # Flipped Vertically 垂直翻轉
im_flip = cv2.flip(im, 0)
#elif method == 'HV':# Flipped Horizontally & Vertically 水平垂直翻轉
# im_flip = cv2.flip(im, -1)
return im_flip
def flip_image(imgs_path, method, anno_new_dir, img_new_dir):
for img_path in imgs_path :
#讀取原圖像
im = cv2.imread(img_path)
flip_img = im_flip(im, method) #翻轉
(H,W,D) = flip_img.shape #得到翻轉後的圖像的高、寬、深度，用於書寫xml
file_name = os.path.basename(os.path.splitext(img_path)[0]) #得到原圖的名稱
ext = os.path.splitext(img_path)[-1] #得到原圖的後綴
#保存翻轉後圖像
new_img_name = '%s_%s%s'%(method, file_name,ext)
cv2.imwrite(os.path.join(img_new_dir, new_img_name), flip_img) #新的命名方式爲H/V+原圖名稱
#讀取anno標籤數據，返回相應的信息
anno = os.path.join(anno_path,'%s.xml'%file_name)
[gts,w,h,d] =readXml(anno)
gt_new = []
for gt in gts:
x1 = gt[0] #xmin
y1 = gt[1] #ymin
x2 = gt[2] #xmax
y2 = gt[3] #ymax
classname = str(gt[4])
if method == 'H':
x1 = w - 1 - x1 #xmax
x2 = w - 1 - x2 #xmin
x1 = limit_value(x1,w)
x2 = limit_value(x2,w)
gt_new.append([x2,y1,x1,y2,classname])
elif method == 'V':
y1 = h - 1 - y1 #ymax
y2 = h - 1 - y2 #ymin
y1 = limit_value(y1,h)
y2 = limit_value(y2,h)
gt_new.append([x1,y2,x2,y1,classname])
writeXml(anno_new_dir, new_img_name, W, H, D, gt_new)
def rotate_image(angles, angle_rad, imgs_path, anno_new_dir, img_new_dir):
j=0 # 計數用
angle_num = len(angles)
for img_path in imgs_path :
#讀取原圖像
im = cv2.imread(img_path)
for i in range(angle_num):
gt_new = []
im_rot = im_rotate(im,angles[i]) #旋轉
(H,W,D) = im_rot.shape #得到旋轉後的圖像的高、寬、深度，用於書寫xml
file_name = os.path.basename(os.path.splitext(img_path)[0]) #得到原圖的名稱
ext = os.path.splitext(img_path)[-1] #得到原圖的後綴
#保存旋轉後圖像
new_img_name = 'P%s_%s%s'%(angles[i],file_name,ext)
cv2.imwrite(os.path.join(img_new_dir, new_img_name),im_rot) #新的命名方式爲P+角度+原圖名稱
#讀取anno標籤數據，返回相應的信息
anno = os.path.join(anno_path,'%s.xml'%file_name)
[gts,w,h,d] =readXml(anno)
#計算旋轉後gt框四點的座標變換
[xc,yc] = [float(w)/2,float(h)/2]
for gt in gts:
#計算左上角點的變換
x1 = (gt[0]-xc)*math.cos(angle_rad[i]) - (yc-gt[1])*math.sin(angle_rad[i]) + xc
if int(x1)<=0 : x1=1.0
if int(x1)>w-1 : x1=w-1
y1 = yc - (gt[0]-xc)*math.sin(angle_rad[i]) - (yc-gt[1])*math.cos(angle_rad[i])
if int(y1)<=0 : y1=1.0
if int(y1)>h-1 : y1=h-1
#計算右上角點的變換
x2 = (gt[2]-xc)*math.cos(angle_rad[i]) - (yc-gt[1])*math.sin(angle_rad[i]) + xc
if int(x2)<=0 : x2=1.0
if int(x2)>w-1 : x2=w-1
y2 = yc - (gt[2]-xc)*math.sin(angle_rad[i]) - (yc-gt[1])*math.cos(angle_rad[i])
if int(y2)<=0 : y2=1.0
if int(y2)>h-1 : y2=h-1
#計算左下角點的變換
x3 = (gt[0]-xc)*math.cos(angle_rad[i]) - (yc-gt[3])*math.sin(angle_rad[i]) + xc
if int(x3)<=0 : x3=1.0
if int(x3)>w-1 : x3=w-1
y3 = yc - (gt[0]-xc)*math.sin(angle_rad[i]) - (yc-gt[3])*math.cos(angle_rad[i])
if int(y3)<=0 : y3=1.0
if int(y3)>h-1 : y3=h-1
#計算右下角點的變換
x4 = (gt[2]-xc)*math.cos(angle_rad[i]) - (yc-gt[3])*math.sin(angle_rad[i]) + xc
if int(x4)<=0 : x4=1.0
if int(x4)>w-1 : x4=w-1
y4 = yc - (gt[2]-xc)*math.sin(angle_rad[i]) - (yc-gt[3])*math.cos(angle_rad[i])
if int(y4)<=0 : y4=1.0
if int(y4)>h-1 : y4=h-1
xmin = min(x1,x2,x3,x4)
xmax = max(x1,x2,x3,x4)
ymin = min(y1,y2,y3,y4)
ymax = max(y1,y2,y3,y4)
#把因爲旋轉導致的特別小的 長線型的去掉
#w_new = xmax-xmin+1
#h_new = ymax-ymin+1
#ratio1 = float(w_new)/h_new
#ratio2 = float(h_new)/w_new
#if(1.0/6.0<ratio1<6 and 1.0/6.0<ratio2<6 and w_new>9 and h_new>9):
classname = str(gt[4])
gt_new.append([xmin,ymin,xmax,ymax,classname])
#寫出新的xml
writeXml(anno_new_dir, new_img_name, W, H, D, gt_new)
j = j+1
if j%100==0 : print ('----%s----'%j)
#寫xml文件，參數中tmp表示路徑，imgname是文件名（沒有尾綴）ps有尾綴也無所謂
def writeXml(tmp, imgname, w, h, d, bboxes):
doc = Document()
#owner
annotation = doc.createElement('annotation')
doc.appendChild(annotation)
#owner
folder = doc.createElement('folder')
annotation.appendChild(folder)
folder_txt = doc.createTextNode("VOC2007")
folder.appendChild(folder_txt)
filename = doc.createElement('filename')
annotation.appendChild(filename)
filename_txt = doc.createTextNode(imgname)
filename.appendChild(filename_txt)
#ones#
source = doc.createElement('source')
annotation.appendChild(source)
database = doc.createElement('database')
source.appendChild(database)
database_txt = doc.createTextNode("My Database")
database.appendChild(database_txt)
annotation_new = doc.createElement('annotation')
source.appendChild(annotation_new)
annotation_new_txt = doc.createTextNode("VOC2007")
annotation_new.appendChild(annotation_new_txt)
image = doc.createElement('image')
source.appendChild(image)
image_txt = doc.createTextNode("flickr")
image.appendChild(image_txt)
#owner
owner = doc.createElement('owner')
annotation.appendChild(owner)
flickrid = doc.createElement('flickrid')
owner.appendChild(flickrid)
flickrid_txt = doc.createTextNode("NULL")
flickrid.appendChild(flickrid_txt)
ow_name = doc.createElement('name')
owner.appendChild(ow_name)
ow_name_txt = doc.createTextNode("idannel")
ow_name.appendChild(ow_name_txt)
#onee#
#twos#
size = doc.createElement('size')
annotation.appendChild(size)
width = doc.createElement('width')
size.appendChild(width)
width_txt = doc.createTextNode(str(w))
width.appendChild(width_txt)
height = doc.createElement('height')
size.appendChild(height)
height_txt = doc.createTextNode(str(h))
height.appendChild(height_txt)
depth = doc.createElement('depth')
size.appendChild(depth)
depth_txt = doc.createTextNode(str(d))
depth.appendChild(depth_txt)
#twoe#
segmented = doc.createElement('segmented')
annotation.appendChild(segmented)
segmented_txt = doc.createTextNode("0")
segmented.appendChild(segmented_txt)
for bbox in bboxes:
#threes#
object_new = doc.createElement("object")
annotation.appendChild(object_new)
name = doc.createElement('name')
object_new.appendChild(name)
name_txt = doc.createTextNode(str(bbox[4]))
name.appendChild(name_txt)
pose = doc.createElement('pose')
object_new.appendChild(pose)
pose_txt = doc.createTextNode("Unspecified")
pose.appendChild(pose_txt)
truncated = doc.createElement('truncated')
object_new.appendChild(truncated)
truncated_txt = doc.createTextNode("0")
truncated.appendChild(truncated_txt)
difficult = doc.createElement('difficult')
object_new.appendChild(difficult)
difficult_txt = doc.createTextNode("0")
difficult.appendChild(difficult_txt)
#threes-1#
bndbox = doc.createElement('bndbox')
object_new.appendChild(bndbox)
xmin = doc.createElement('xmin')
bndbox.appendChild(xmin)
xmin_txt = doc.createTextNode(str(bbox[0]))
xmin.appendChild(xmin_txt)
ymin = doc.createElement('ymin')
bndbox.appendChild(ymin)
ymin_txt = doc.createTextNode(str(bbox[1]))
ymin.appendChild(ymin_txt)
xmax = doc.createElement('xmax')
bndbox.appendChild(xmax)
xmax_txt = doc.createTextNode(str(bbox[2]))
xmax.appendChild(xmax_txt)
ymax = doc.createElement('ymax')
bndbox.appendChild(ymax)
ymax_txt = doc.createTextNode(str(bbox[3]))
ymax.appendChild(ymax_txt)
print(bbox[0],bbox[1],bbox[2],bbox[3])
xmlname = os.path.splitext(imgname)[0]
tempfile = tmp +"/%s.xml"%xmlname
with open(tempfile, 'wb') as f:
f.write(doc.toprettyxml(indent='\t', encoding='utf-8'))
return
if __name__ == '__main__':
#數據路徑
root = '/home/yantianwang/clone/haha'
img_dir = root + '/images'
anno_path = root + '/xml'
imgs_path=GetFileFromThisRootDir(img_dir)#返回每一張原圖的路徑
AUG = 'Flip' #數據擴增的方式 Rotate代表旋轉，Flip表示翻轉
#存儲新的anno位置
anno_new_dir = os.path.join(root, AUG , 'xml')
if not os.path.isdir(anno_new_dir):
os.makedirs(anno_new_dir)
#擴增後圖片保存的位置
img_new_dir = os.path.join(root, AUG , 'images')
if not os.path.isdir(img_new_dir):
os.makedirs(img_new_dir)
if AUG == 'Rotate':
#旋轉角的大小，正數表示逆時針旋轉
angles = [5,90,180,270,355]#角度im_rotate用到的是角度制
angle_rad = [angle*math.pi/180.0 for angle in angles] #cos三角函數裏要用到弧度制的
#開始旋轉
rotate_image(angles, angle_rad, imgs_path, anno_new_dir, img_new_dir)
elif AUG == 'Flip':
method = 'H'
flip_image(imgs_path, method, anno_new_dir, img_new_dir)
標題OBB格式的數據擴增
該格式下目標被標註爲x0，y0，x1，y1，x2，y2，x3，y3
import os
import cv2
import xml.dom.minidom
from xml.dom.minidom import Document
import math
import pdb
#獲取路徑下所有文件的完整路徑，用於讀取文件用
def GetFileFromThisRootDir(dir,ext = None):
allfiles = []
needExtFilter = (ext != None)
for root,dirs,files in os.walk(dir):
for filespath in files:
filepath = os.path.join(root, filespath)
extension = os.path.splitext(filepath)[1][1:]
if needExtFilter and extension in ext:
allfiles.append(filepath)
elif not needExtFilter:
allfiles.append(filepath)
return allfiles
def limit_value(a,b):
if a<1:
a = 1
if a>b:
a = b-1
return a
#讀取xml文件，xmlfile參數表示xml的路徑
def readXml(xmlfile):
DomTree = xml.dom.minidom.parse(xmlfile)
annotation = DomTree.documentElement
sizelist = annotation.getElementsByTagName('size') #[<DOM Element: filename at 0x381f788>]
heights = sizelist[0].getElementsByTagName('height')
height = int(heights[0].childNodes[0].data)
widths =sizelist[0].getElementsByTagName('width')
width = int(widths[0].childNodes[0].data)
depths = sizelist[0].getElementsByTagName('depth')
depth = int(depths[0].childNodes[0].data)
objectlist = annotation.getElementsByTagName('object')
bboxes = []
for objects in objectlist:
namelist = objects.getElementsByTagName('name')
class_label = namelist[0].childNodes[0].data
bndbox = objects.getElementsByTagName('bndbox')[0]
x0_list = bndbox.getElementsByTagName('x0')
x0 = int(float(x0_list[0].childNodes[0].data))
y0_list = bndbox.getElementsByTagName('y0')
y0 = int(float(y0_list[0].childNodes[0].data))
x1_list = bndbox.getElementsByTagName('x1')
x1 = int(float(x1_list[0].childNodes[0].data))
y1_list = bndbox.getElementsByTagName('y1')
y1 = int(float(y1_list[0].childNodes[0].data))
x2_list = bndbox.getElementsByTagName('x2')
x2 = int(float(x2_list[0].childNodes[0].data))
y2_list = bndbox.getElementsByTagName('y2')
y2 = int(float(y2_list[0].childNodes[0].data))
x3_list = bndbox.getElementsByTagName('x3')
x3 = int(float(x3_list[0].childNodes[0].data))
y3_list = bndbox.getElementsByTagName('y3')
y3 = int(float(y3_list[0].childNodes[0].data))
#這裏我box的格式【xmin，ymin，xmax，ymax，classname】
bbox = [x0,y0,x1,y1,x2,y2,x3,y3,class_label]
bboxes.append(bbox)
return bboxes,width,height,depth
#圖像旋轉用，裏面的angle是角度制的
def im_rotate(im,angle,center = None,scale = 1.0):
h,w = im.shape[:2]
if center is None:
center = (w/2,h/2)
M = cv2.getRotationMatrix2D(center,angle,scale)
im_rot = cv2.warpAffine(im,M,(w,h))
return im_rot
def im_flip(im, method = 'H'): #翻轉圖像
if method == 'H': # Flipped Horizontally 水平翻轉
im_flip = cv2.flip(im, 1)
elif method == 'V': # Flipped Vertically 垂直翻轉
im_flip = cv2.flip(im, 0)
#elif method == 'HV':# Flipped Horizontally & Vertically 水平垂直翻轉
# im_flip = cv2.flip(im, -1)
return im_flip
def flip_image(imgs_path, method, anno_new_dir, img_new_dir):
for img_path in imgs_path :
#讀取原圖像
im = cv2.imread(img_path)
flip_img = im_flip(im, method) #翻轉
(H,W,D) = flip_img.shape #得到翻轉後的圖像的高、寬、深度，用於書寫xml
file_name = os.path.basename(os.path.splitext(img_path)[0]) #得到原圖的名稱
ext = os.path.splitext(img_path)[-1] #得到原圖的後綴
#保存翻轉後圖像
new_img_name = '%s_%s%s'%(method, file_name,ext)
cv2.imwrite(os.path.join(img_new_dir, new_img_name), flip_img) #新的命名方式爲H/V+原圖名稱
#讀取anno標籤數據，返回相應的信息
anno = os.path.join(anno_path,'%s.xml'%file_name)
[gts,w,h,d] =readXml(anno)
gt_new = []
for gt in gts:
x0 = gt[0]
y0 = gt[1]
x1 = gt[2]
y1 = gt[3]
x2 = gt[4]
y2 = gt[5]
x3 = gt[6]
y3 = gt[7]
classname = gt[-1]
if method == 'H':
x0 = w - x0
x1 = w - x1
x2 = w - x2
x3 = w - x3
x0 = limit_value(x0,w)
x1 = limit_value(x1,w)
x2 = limit_value(x2,w)
x3 = limit_value(x3,w)
elif method == 'V':
y0 = h - y0
y1 = h - y1
y2 = h - y2
y3 = h - y3
y0 = limit_value(y0,h)
y1 = limit_value(y1,h)
y2 = limit_value(y2,h)
y3 = limit_value(y3,h)
gt_new.append([x0,y0,x1,y1,x2,y2,x3,y3,classname])
writeXml(anno_new_dir, new_img_name, W, H, D, gt_new)
def rotate_image(angles, angle_rad, imgs_path, anno_new_dir, img_new_dir):
j=0 # 計數用
angle_num = len(angles)
for img_path in imgs_path :
#讀取原圖像
im = cv2.imread(img_path)
for i in range(angle_num):
gt_new = []
im_rot = im_rotate(im,angles[i]) #旋轉
(H,W,D) = im_rot.shape #得到旋轉後的圖像的高、寬、深度，用於書寫xml
file_name = os.path.basename(os.path.splitext(img_path)[0]) #得到原圖的名稱
ext = os.path.splitext(img_path)[-1] #得到原圖的後綴
#保存旋轉後圖像
new_img_name = 'P%s_%s%s'%(angles[i],file_name,ext)
cv2.imwrite(os.path.join(img_new_dir, new_img_name),im_rot) #新的命名方式爲P+角度+原圖名稱
#讀取anno標籤數據，返回相應的信息
anno = os.path.join(anno_path,'%s.xml'%file_name)
[gts,w,h,d] =readXml(anno)
#計算旋轉後gt框四點的座標變換
[xc,yc] = [float(w)/2,float(h)/2]
for gt in gts:
#計算左上角點的變換
x1 = (gt[0]-xc)*math.cos(angle_rad[i]) - (yc-gt[1])*math.sin(angle_rad[i]) + xc
if int(x1)<=0 : x1=1
if int(x1)>w-1 : x1=w-1
y1 = yc - (gt[0]-xc)*math.sin(angle_rad[i]) - (yc-gt[1])*math.cos(angle_rad[i])
if int(y1)<=0 : y1=1.0
if int(y1)>h-1 : y1=h-1
#計算右上角點的變換
x2 = (gt[2]-xc)*math.cos(angle_rad[i]) - (yc-gt[3])*math.sin(angle_rad[i]) + xc
if int(x2)<=0 : x2=1
if int(x2)>w-1 : x2=w-1
y2 = yc - (gt[2]-xc)*math.sin(angle_rad[i]) - (yc-gt[3])*math.cos(angle_rad[i])
if int(y2)<=0 : y2=1
if int(y2)>h-1 : y2=h-1
#計算左下角點的變換
x3 = (gt[4]-xc)*math.cos(angle_rad[i]) - (yc-gt[5])*math.sin(angle_rad[i]) + xc
if int(x3)<=0 : x3=1
if int(x3)>w-1 : x3=w-1
y3 = yc - (gt[4]-xc)*math.sin(angle_rad[i]) - (yc-gt[5])*math.cos(angle_rad[i])
if int(y3)<=0 : y3=1
if int(y3)>h-1 : y3=h-1
#計算右下角點的變換
x4 = (gt[6]-xc)*math.cos(angle_rad[i]) - (yc-gt[7])*math.sin(angle_rad[i]) + xc
if int(x4)<=0 : x4=1
if int(x4)>w-1 : x4=w-1
y4 = yc - (gt[6]-xc)*math.sin(angle_rad[i]) - (yc-gt[7])*math.cos(angle_rad[i])
if int(y4)<=0 : y4=1
if int(y4)>h-1 : y4=h-1
#xmin = min(x1,x2,x3,x4)
#xmax = max(x1,x2,x3,x4)
#ymin = min(y1,y2,y3,y4)
#ymax = max(y1,y2,y3,y4)
#把因爲旋轉導致的特別小的 長線型的去掉
#w_new = xmax-xmin+1
#h_new = ymax-ymin+1
#ratio1 = float(w_new)/h_new
#ratio2 = float(h_new)/w_new
#if(1.0/6.0<ratio1<6 and 1.0/6.0<ratio2<6 and w_new>9 and h_new>9):
classname = gt[-1]
gt_new.append([int(x1),int(y1),int(x2),int(y2),int(x3),int(y3),int(x4),int(y4),classname])
#寫出新的xml
writeXml(anno_new_dir, new_img_name, W, H, D, gt_new)
j = j+1
if j%100==0 : print ('----%s----'%j)
#寫xml文件，參數中tmp表示路徑，imgname是文件名（沒有尾綴）ps有尾綴也無所謂
def writeXml(tmp, imgname, w, h, d, bboxes):
doc = Document()
#owner
annotation = doc.createElement('annotation')
doc.appendChild(annotation)
#owner
folder = doc.createElement('folder')
annotation.appendChild(folder)
folder_txt = doc.createTextNode("VOC2007")
folder.appendChild(folder_txt)
filename = doc.createElement('filename')
annotation.appendChild(filename)
filename_txt = doc.createTextNode(imgname)
filename.appendChild(filename_txt)
#ones#
source = doc.createElement('source')
annotation.appendChild(source)
database = doc.createElement('database')
source.appendChild(database)
database_txt = doc.createTextNode("My Database")
database.appendChild(database_txt)
annotation_new = doc.createElement('annotation')
source.appendChild(annotation_new)
annotation_new_txt = doc.createTextNode("VOC2007")
annotation_new.appendChild(annotation_new_txt)
image = doc.createElement('image')
source.appendChild(image)
image_txt = doc.createTextNode("flickr")
image.appendChild(image_txt)
#owner
owner = doc.createElement('owner')
annotation.appendChild(owner)
flickrid = doc.createElement('flickrid')
owner.appendChild(flickrid)
flickrid_txt = doc.createTextNode("NULL")
flickrid.appendChild(flickrid_txt)
ow_name = doc.createElement('name')
owner.appendChild(ow_name)
ow_name_txt = doc.createTextNode("idannel")
ow_name.appendChild(ow_name_txt)
#onee#
#twos#
size = doc.createElement('size')
annotation.appendChild(size)
width = doc.createElement('width')
size.appendChild(width)
width_txt = doc.createTextNode(str(w))
width.appendChild(width_txt)
height = doc.createElement('height')
size.appendChild(height)
height_txt = doc.createTextNode(str(h))
height.appendChild(height_txt)
depth = doc.createElement('depth')
size.appendChild(depth)
depth_txt = doc.createTextNode(str(d))
depth.appendChild(depth_txt)
#twoe#
segmented = doc.createElement('segmented')
annotation.appendChild(segmented)
segmented_txt = doc.createTextNode("0")
segmented.appendChild(segmented_txt)
for bbox in bboxes:
#threes#
object_new = doc.createElement("object")
annotation.appendChild(object_new)
name = doc.createElement('name')
object_new.appendChild(name)
name_txt = doc.createTextNode(bbox[-1])
name.appendChild(name_txt)
pose = doc.createElement('pose')
object_new.appendChild(pose)
pose_txt = doc.createTextNode("Unspecified")
pose.appendChild(pose_txt)
truncated = doc.createElement('truncated')
object_new.appendChild(truncated)
truncated_txt = doc.createTextNode("0")
truncated.appendChild(truncated_txt)
difficult = doc.createElement('difficult')
object_new.appendChild(difficult)
difficult_txt = doc.createTextNode("0")
difficult.appendChild(difficult_txt)
#threes-1#
bndbox = doc.createElement('bndbox')
object_new.appendChild(bndbox)
x0 = doc.createElement('x0')
bndbox.appendChild(x0)
x0_txt = doc.createTextNode(str(bbox[0]))
x0.appendChild(x0_txt)
y0 = doc.createElement('y0')
bndbox.appendChild(y0)
y0_txt = doc.createTextNode(str(bbox[1]))
y0.appendChild(y0_txt)
x1 = doc.createElement('x1')
bndbox.appendChild(x1)
x1_txt = doc.createTextNode(str(bbox[2]))
x1.appendChild(x1_txt)
y1 = doc.createElement('y1')
bndbox.appendChild(y1)
y1_txt = doc.createTextNode(str(bbox[3]))
y1.appendChild(y1_txt)
x2 = doc.createElement('x2')
bndbox.appendChild(x2)
x2_txt = doc.createTextNode(str(bbox[4]))
x2.appendChild(x2_txt)
y2 = doc.createElement('y2')
bndbox.appendChild(y2)
y2_txt = doc.createTextNode(str(bbox[5]))
y2.appendChild(y2_txt)
x3 = doc.createElement('x3')
bndbox.appendChild(x3)
x3_txt = doc.createTextNode(str(bbox[6]))
x3.appendChild(x3_txt)
y3 = doc.createElement('y3')
bndbox.appendChild(y3)
y3_txt = doc.createTextNode(str(bbox[7]))
y3.appendChild(y3_txt)
#print(float(int(bbox[0])),float(int(bbox[1])),float(int(bbox[2])),float(int(bbox[3])))
xmlname = os.path.splitext(imgname)[0]
tempfile = tmp +"/%s.xml"%xmlname
with open(tempfile, 'wb') as f:
f.write(doc.toprettyxml(indent='\t', encoding='utf-8'))
return
if __name__ == '__main__':
#數據路徑
root = '/home/yantianwang/lala/20190225/oldship/examplesplit'
img_dir = root + '/images'
anno_path = root + '/obbxml'
imgs_path=GetFileFromThisRootDir(img_dir)#返回每一張原圖的路徑
AUG = 'Flip' #數據擴增的方式 Rotate代表旋轉，Flip表示翻轉
#存儲新的anno位置
anno_new_dir = os.path.join(root, AUG , 'NewAnnotations')
#anno_new_path = root + '/NewAnnotations'
if not os.path.isdir(anno_new_dir):
os.makedirs(anno_new_dir)
#擴增後圖片保存的位置
img_new_dir = os.path.join(root, AUG , 'Newimage')
if not os.path.isdir(img_new_dir):
os.makedirs(img_new_dir)
if AUG == 'Rotate':
#旋轉角的大小，正數表示逆時針旋轉
angles = [5,90,180,270,355]#角度im_rotate用到的是角度制
angle_rad = [angle*math.pi/180.0 for angle in angles] #cos三角函數裏要用到弧度制的
#開始旋轉
rotate_image(angles, angle_rad, imgs_path, anno_new_dir, img_new_dir)
elif AUG == 'Flip':
method = 'H'
flip_image(imgs_path, method, anno_new_dir, img_new_dir)