将wierface标注转换为VOC格式

widerface是包含了3万多张总计近40万张人脸的人脸检测库，里面包含了大大小小各式各样的人脸，是不可多得的素材。

请将下面的代码保存至widerface.py，并至于下图所示的eval_tools文件夹下，其他的文件结构一并如图所示。

Update:

由于widerface里包含很多小脸，用SSD训练不一定能收敛，此外SSD要求输入为方形，不然会挤压图片造成变形，因此需要对此做些处理.

[cpp] view plain copy

print?

[python] view plain copy

print?

[python] view plain copy

print?

1. import
2. import
3. fromimport
4. rootdir=”../”
5. convet2yoloformat=True
6. convert2vocformat=True
7. resized_dim=(4848
9. #最小取20大小的脸，并且补齐
10. minsize2select=20
11. usepadding=True
13. datasetprefix=”/home/yanhe/data/widerface”#
14. def
15. ”/WIDER_train/images”
16. ”/wider_face_split/wider_face_train_bbx_gt.txt”
17. 0
18. ’r’
21. while(True
22. 1
23. if
24. break
25. print
26. ”/”
29. forin
32. 0:4
33. if(int(line[3])<=0or2])<=0
34. continue
35. 0]),int(line[1]),int(line[2]),int(line[3
36. 1]):int(line[1])+int(line[3]),int(line[0]):int(line[0])+int(line[2
39. 1
40. 0]),int(line[1])),(int(line[0])+int(line[2]),int(line[1])+int(line[3])),(255,0,0
41. #cv2.imshow(“img”,img)
42. #cv2.waitKey(1)
43. 1
46. ’train.h5’,‘w’
47. ’data’
48. ’label’
50. def
51. ’train.h5’,‘r’
53. ’data’
54. forin
56. ”img”
57. 1
60. def“train”
61. ”/WIDER_”+img_set+“/images”
62. ”/wider_face_split/wider_face_”+img_set+“_bbx_gt.txt”
63. ”/images”
64. ”/Annotations”
65. ”/labels”
66. ifnot
68. if
69. ifnot
71. if
72. ifnot
74. 0
75. ’r’
76. while(True
77. 1
78. if
79. break
80. ”\r”+str(index)+“:”+filename+“\t\t\t”
82. ”/”
84. ifnot
85. break
86. 0
87. 1
89. 1
90. 1
91. 1
92. 1
93. 0
97. forin
100. 0:4
101. if(int(line[3])<=0or2])<=0
102. continue
103. 0
104. 1
105. 2
106. 3
110. #face=img[x:x2,y:y2]
111. ifand
113. 0,255,0
114. #maxl=max(width,height)
115. #x3=(int)(x+(width-maxl)*0.5)
116. #y3=(int)(y+(height-maxl)*0.5)
117. #x4=(int)(x3+maxl)
118. #y4=(int)(y3+maxl)
119. #cv2.rectangle(img,(x3,y3),(x4,y4),(255,0,0))
120. else
121. 0,0,255
122. ”/”,“_”
123. if0
124. print
125. continue
126. ”/”
127. if
128. 0
129. 1
130. ”/”
131. 3]+“txt”
132. ’w’
133. forin
135. 0]+bbox[2]*0.5
136. 1]+bbox[3]*0.5
137. 2]*1.0
138. 3]*1.0
139. +str(xcenter)++str(ycenter)++str(wr)++str(hr)+“\n”
142. if
143. ”/”
144. 3]+“xml”
146. ’annotation’
148. ’folder’
149. ’widerface’
152. ’filename’
156. ’source’
158. ’database’
161. ’annotation’
164. ’image’
165. ’flickr’
167. ’flickrid’
168. ’-1’
170. ’owner’
172. ’flickrid’
173. ’yanyu’
175. ’name’
176. ’yanyu’
178. ’size’
180. ’width’
181. 1
182. ’height’
183. 0
184. ’depth’
185. 2
189. ’segmented’
190. ’0’
192. forin
194. ’object’
196. ’name’
197. ’face’
199. ’pose’
200. ’Unspecified’
202. ’truncated’
203. ’1’
205. ’difficult’
206. ’0’
208. ’bndbox’
210. ’xmin’
211. 0
213. ’ymin’
214. 1
216. ’xmax’
217. 0]+bbox[2
219. ’ymax’
220. 1]+bbox[3
222. ”w”
223. ”
225. #cv2.imshow(“img”,showimg)
226. #cv2.waitKey()
227. 1
229. def“train”
230. ”/wider_face_split/wider_face_”+img_set+“_bbx_gt.txt”
231. ”/”+img_set+“.txt”,“w”
232. ’r’
233. while(True
234. 1
235. if
236. break
237. ”/”,“_”
238. ”/images/”
239. ’\n’
241. forin
245. def“train”
246. ifnot“/ImageSets”
247. ”/ImageSets”
248. ifnot“/ImageSets/Main”
249. ”/ImageSets/Main”
250. ”/wider_face_split/wider_face_”+img_set+“_bbx_gt.txt”
251. ”/ImageSets/Main/”+img_set+“.txt”,‘w’
252. ’r’
253. while(True
254. 1
255. if
256. break
257. ”/”,“_”
258. 4
259. ’\n’
261. forin
265. def
266. ”train”,“val”
267. forin
272. if“__main__”
274. ”/”+“train.txt”,rootdir+“/”+“trainval.txt”
275. ”/”+“val.txt”,rootdir+“/”+“test.txt”
276. ”/ImageSets/Main/”+“train.txt”,rootdir+“/ImageSets/Main/”+“trainval.txt”
277. ”/ImageSets/Main/”+“val.txt”,rootdir+“/ImageSets/Main/”+“test.txt”

import os,h5py,cv2,sys,shutil import numpy as np from xml.dom.minidom import Document rootdir="../" convet2yoloformat=True convert2vocformat=True resized_dim=(48, 48)      

minsize2select=20
usepadding=True

datasetprefix="/home/yanhe/data/widerface"#
def gen_hdf5():
imgdir=rootdir+"/WIDER_train/images"
gtfilepath=rootdir+"/wider_face_split/wider_face_train_bbx_gt.txt"
index =0
with open(gtfilepath,'r') as gtfile:
faces=[]
labels=[]
while(True ):#and len(faces)<10
imgpath=gtfile.readline()[:-1]
if(imgpath==""):
break;
print index,imgpath
img=cv2.imread(imgdir+"/"+imgpath)
numbbox=int(gtfile.readline())
bbox=[]
for i in range(numbbox):
line=gtfile.readline()
line=line.split()
line=line[0:4]
if(int(line[3])<=0 or int(line[2])<=0):
continue
bbox=(int(line[0]),int(line[1]),int(line[2]),int(line[3]))
face=img[int(line[1]):int(line[1])+int(line[3]),int(line[0]):int(line[0])+int(line[2])]
face=cv2.resize(face, resized_dim)
faces.append(face)
labels.append(1)
cv2.rectangle(img,(int(line[0]),int(line[1])),(int(line[0])+int(line[2]),int(line[1])+int(line[3])),(255,0,0))
#cv2.imshow("img",img)
#cv2.waitKey(1)
index=index+1
faces=np.asarray(faces)
labels=np.asarray(labels)
f=h5py.File('train.h5','w')
f['data']=faces.astype(np.float32)
f['label']=labels.astype(np.float32)
f.close()
def viewginhdf5():
f = h5py.File('train.h5','r')
f.keys()
faces=f['data'][:]
for face in faces:
face=face.astype(np.uint8)
cv2.imshow("img",face)
cv2.waitKey(1)
f.close()

def convertimgset(img_set="train"):
imgdir=rootdir+"/WIDER_"+img_set+"/images"
gtfilepath=rootdir+"/wider_face_split/wider_face_"+img_set+"_bbx_gt.txt"
imagesdir=rootdir+"/images"
vocannotationdir=rootdir+"/Annotations"
labelsdir=rootdir+"/labels"
if not os.path.exists(imagesdir):
os.mkdir(imagesdir)
if convet2yoloformat:
if not os.path.exists(labelsdir):
os.mkdir(labelsdir)
if convert2vocformat:
if not os.path.exists(vocannotationdir):
os.mkdir(vocannotationdir)
index=0
with open(gtfilepath,'r') as gtfile:
while(True ):#and len(faces)<10
filename=gtfile.readline()[:-1]
if(filename==""):
break;
sys.stdout.write("\r"+str(index)+":"+filename+"\t\t\t")
sys.stdout.flush()
imgpath=imgdir+"/"+filename
img=cv2.imread(imgpath)
if not img.data:
break;
imgheight=img.shape[0]
imgwidth=img.shape[1]
maxl=max(imgheight,imgwidth)
paddingleft=(maxl-imgwidth)>>1
paddingright=(maxl-imgwidth)>>1
paddingbottom=(maxl-imgheight)>>1
paddingtop=(maxl-imgheight)>>1
saveimg=cv2.copyMakeBorder(img,paddingtop,paddingbottom,paddingleft,paddingright,cv2.BORDER_CONSTANT,value=0)
showimg=saveimg.copy()
numbbox=int(gtfile.readline())
bboxes=[]
for i in range(numbbox):
line=gtfile.readline()
line=line.split()
line=line[0:4]
if(int(line[3])<=0 or int(line[2])<=0):
continue
x=int(line[0])+paddingleft
y=int(line[1])+paddingtop
width=int(line[2])
height=int(line[3])
bbox=(x,y,width,height)
x2=x+width
y2=y+height
#face=img[x:x2,y:y2]
if width>=minsize2select and height>=minsize2select:
bboxes.append(bbox)
cv2.rectangle(showimg,(x,y),(x2,y2),(0,255,0))
#maxl=max(width,height)
#x3=(int)(x+(width-maxl)*0.5)
#y3=(int)(y+(height-maxl)*0.5)
#x4=(int)(x3+maxl)
#y4=(int)(y3+maxl)
#cv2.rectangle(img,(x3,y3),(x4,y4),(255,0,0))
else:
cv2.rectangle(showimg,(x,y),(x2,y2),(0,0,255))
filename=filename.replace("/","_")
if len(bboxes)==0:
print "warrning: no face"
continue
cv2.imwrite(imagesdir+"/"+filename,saveimg)
if convet2yoloformat:
height=saveimg.shape[0]
width=saveimg.shape[1]
txtpath=labelsdir+"/"+filename
txtpath=txtpath[:-3]+"txt"
ftxt=open(txtpath,'w')
for i in range(len(bboxes)):
bbox=bboxes[i]
xcenter=(bbox[0]+bbox[2]*0.5)/width
ycenter=(bbox[1]+bbox[3]*0.5)/height
wr=bbox[2]*1.0/width
hr=bbox[3]*1.0/height
txtline="0 "+str(xcenter)+" "+str(ycenter)+" "+str(wr)+" "+str(hr)+"\n"
ftxt.write(txtline)
ftxt.close()
if convert2vocformat:
xmlpath=vocannotationdir+"/"+filename
xmlpath=xmlpath[:-3]+"xml"
doc = Document()
annotation = doc.createElement('annotation')
doc.appendChild(annotation)
folder = doc.createElement('folder')
folder_name = doc.createTextNode('widerface')
folder.appendChild(folder_name)
annotation.appendChild(folder)
filenamenode = doc.createElement('filename')
filename_name = doc.createTextNode(filename)
filenamenode.appendChild(filename_name)
annotation.appendChild(filenamenode)
source = doc.createElement('source')
annotation.appendChild(source)
database = doc.createElement('database')
database.appendChild(doc.createTextNode('wider face Database'))
source.appendChild(database)
annotation_s = doc.createElement('annotation')
annotation_s.appendChild(doc.createTextNode('PASCAL VOC2007'))
source.appendChild(annotation_s)
image = doc.createElement('image')
image.appendChild(doc.createTextNode('flickr'))
source.appendChild(image)
flickrid = doc.createElement('flickrid')
flickrid.appendChild(doc.createTextNode('-1'))
source.appendChild(flickrid)
owner = doc.createElement('owner')
annotation.appendChild(owner)
flickrid_o = doc.createElement('flickrid')
flickrid_o.appendChild(doc.createTextNode('yanyu'))
owner.appendChild(flickrid_o)
name_o = doc.createElement('name')
name_o.appendChild(doc.createTextNode('yanyu'))
owner.appendChild(name_o)
size = doc.createElement('size')
annotation.appendChild(size)
width = doc.createElement('width')
width.appendChild(doc.createTextNode(str(saveimg.shape[1])))
height = doc.createElement('height')
height.appendChild(doc.createTextNode(str(saveimg.shape[0])))
depth = doc.createElement('depth')
depth.appendChild(doc.createTextNode(str(saveimg.shape[2])))
size.appendChild(width)
size.appendChild(height)
size.appendChild(depth)
segmented = doc.createElement('segmented')
segmented.appendChild(doc.createTextNode('0'))
annotation.appendChild(segmented)
for i in range(len(bboxes)):
bbox=bboxes[i]
objects = doc.createElement('object')
annotation.appendChild(objects)
object_name = doc.createElement('name')
object_name.appendChild(doc.createTextNode('face'))
objects.appendChild(object_name)
pose = doc.createElement('pose')
pose.appendChild(doc.createTextNode('Unspecified'))
objects.appendChild(pose)
truncated = doc.createElement('truncated')
truncated.appendChild(doc.createTextNode('1'))
objects.appendChild(truncated)
difficult = doc.createElement('difficult')
difficult.appendChild(doc.createTextNode('0'))
objects.appendChild(difficult)
bndbox = doc.createElement('bndbox')
objects.appendChild(bndbox)
xmin = doc.createElement('xmin')
xmin.appendChild(doc.createTextNode(str(bbox[0])))
bndbox.appendChild(xmin)
ymin = doc.createElement('ymin')
ymin.appendChild(doc.createTextNode(str(bbox[1])))
bndbox.appendChild(ymin)
xmax = doc.createElement('xmax')
xmax.appendChild(doc.createTextNode(str(bbox[0]+bbox[2])))
bndbox.appendChild(xmax)
ymax = doc.createElement('ymax')
ymax.appendChild(doc.createTextNode(str(bbox[1]+bbox[3])))
bndbox.appendChild(ymax)
f=open(xmlpath,"w")
f.write(doc.toprettyxml(indent = ''))
f.close()
#cv2.imshow("img",showimg)
#cv2.waitKey()
index=index+1

def generatetxt(img_set="train"):
gtfilepath=rootdir+"/wider_face_split/wider_face_"+img_set+"_bbx_gt.txt"
f=open(rootdir+"/"+img_set+".txt","w")
with open(gtfilepath,'r') as gtfile:
while(True ):#and len(faces)<10
filename=gtfile.readline()[:-1]
if(filename==""):
break;
filename=filename.replace("/","_")
imgfilepath=datasetprefix+"/images/"+filename
f.write(imgfilepath+'\n')
numbbox=int(gtfile.readline())
for i in range(numbbox):
line=gtfile.readline()
f.close()

def generatevocsets(img_set="train"):
if not os.path.exists(rootdir+"/ImageSets"):
os.mkdir(rootdir+"/ImageSets")
if not os.path.exists(rootdir+"/ImageSets/Main"):
os.mkdir(rootdir+"/ImageSets/Main")
gtfilepath=rootdir+"/wider_face_split/wider_face_"+img_set+"_bbx_gt.txt"
f=open(rootdir+"/ImageSets/Main/"+img_set+".txt",'w')
with open(gtfilepath,'r') as gtfile:
while(True ):#and len(faces)<10
filename=gtfile.readline()[:-1]
if(filename==""):
break;
filename=filename.replace("/","_")
imgfilepath=filename[:-4]
f.write(imgfilepath+'\n')
numbbox=int(gtfile.readline())
for i in range(numbbox):
line=gtfile.readline()
f.close()

def convertdataset():
img_sets=["train","val"]
for img_set in img_sets:
convertimgset(img_set)
generatetxt(img_set)
generatevocsets(img_set)

if name=="main":
convertdataset()
shutil.move(rootdir+"/"+"train.txt",rootdir+"/"+"trainval.txt")
shutil.move(rootdir+"/"+"val.txt",rootdir+"/"+"test.txt")
shutil.move(rootdir+"/ImageSets/Main/"+"train.txt",rootdir+"/ImageSets/Main/"+"trainval.txt")
shutil.move(rootdir+"/ImageSets/Main/"+"val.txt",rootdir+"/ImageSets/Main/"+"test.txt")

如果没有时间自己转换，也可以下载已经转换好的文件，百度网盘，密码：t46t

widerface是包含了3万多张总计近40万张人脸的人脸检测库，里面包含了大大小小各式各样的人脸，是不可多得的素材。

请将下面的代码保存至widerface.py，并至于下图所示的eval_tools文件夹下，其他的文件结构一并如图所示。

Update:

由于widerface里包含很多小脸，用SSD训练不一定能收敛，此外SSD要求输入为方形，不然会挤压图片造成变形，因此需要对此做些处理.

[cpp] view plain copy

print?

[python] view plain copy

print?

[python] view plain copy

print?

1. import
2. import
3. fromimport
4. rootdir=”../”
5. convet2yoloformat=True
6. convert2vocformat=True
7. resized_dim=(4848
9. #最小取20大小的脸，并且补齐
10. minsize2select=20
11. usepadding=True
13. datasetprefix=”/home/yanhe/data/widerface”#
14. def
15. ”/WIDER_train/images”
16. ”/wider_face_split/wider_face_train_bbx_gt.txt”
17. 0
18. ’r’
21. while(True
22. 1
23. if
24. break
25. print
26. ”/”
29. forin
32. 0:4
33. if(int(line[3])<=0or2])<=0
34. continue
35. 0]),int(line[1]),int(line[2]),int(line[3
36. 1]):int(line[1])+int(line[3]),int(line[0]):int(line[0])+int(line[2
39. 1
40. 0]),int(line[1])),(int(line[0])+int(line[2]),int(line[1])+int(line[3])),(255,0,0
41. #cv2.imshow(“img”,img)
42. #cv2.waitKey(1)
43. 1
46. ’train.h5’,‘w’
47. ’data’
48. ’label’
50. def
51. ’train.h5’,‘r’
53. ’data’
54. forin
56. ”img”
57. 1
60. def“train”
61. ”/WIDER_”+img_set+“/images”
62. ”/wider_face_split/wider_face_”+img_set+“_bbx_gt.txt”
63. ”/images”
64. ”/Annotations”
65. ”/labels”
66. ifnot
68. if
69. ifnot
71. if
72. ifnot
74. 0
75. ’r’
76. while(True
77. 1
78. if
79. break
80. ”\r”+str(index)+“:”+filename+“\t\t\t”
82. ”/”
84. ifnot
85. break
86. 0
87. 1
89. 1
90. 1
91. 1
92. 1
93. 0
97. forin
100. 0:4
101. if(int(line[3])<=0or2])<=0
102. continue
103. 0
104. 1
105. 2
106. 3
110. #face=img[x:x2,y:y2]
111. ifand
113. 0,255,0
114. #maxl=max(width,height)
115. #x3=(int)(x+(width-maxl)*0.5)
116. #y3=(int)(y+(height-maxl)*0.5)
117. #x4=(int)(x3+maxl)
118. #y4=(int)(y3+maxl)
119. #cv2.rectangle(img,(x3,y3),(x4,y4),(255,0,0))
120. else
121. 0,0,255
122. ”/”,“_”
123. if0
124. print
125. continue
126. ”/”
127. if
128. 0
129. 1
130. ”/”
131. 3]+“txt”
132. ’w’
133. forin
135. 0]+bbox[2]*0.5
136. 1]+bbox[3]*0.5
137. 2]*1.0
138. 3]*1.0
139. +str(xcenter)++str(ycenter)++str(wr)++str(hr)+“\n”
142. if
143. ”/”
144. 3]+“xml”
146. ’annotation’
148. ’folder’
149. ’widerface’
152. ’filename’
156. ’source’
158. ’database’
161. ’annotation’
164. ’image’
165. ’flickr’
167. ’flickrid’
168. ’-1’
170. ’owner’
172. ’flickrid’
173. ’yanyu’
175. ’name’
176. ’yanyu’
178. ’size’
180. ’width’
181. 1
182. ’height’
183. 0
184. ’depth’
185. 2
189. ’segmented’
190. ’0’
192. forin
194. ’object’
196. ’name’
197. ’face’
199. ’pose’
200. ’Unspecified’
202. ’truncated’
203. ’1’
205. ’difficult’
206. ’0’
208. ’bndbox’
210. ’xmin’
211. 0
213. ’ymin’
214. 1
216. ’xmax’
217. 0]+bbox[2
219. ’ymax’
220. 1]+bbox[3
222. ”w”
223. ”
225. #cv2.imshow(“img”,showimg)
226. #cv2.waitKey()
227. 1
229. def“train”
230. ”/wider_face_split/wider_face_”+img_set+“_bbx_gt.txt”
231. ”/”+img_set+“.txt”,“w”
232. ’r’
233. while(True
234. 1
235. if
236. break
237. ”/”,“_”
238. ”/images/”
239. ’\n’
241. forin
245. def“train”
246. ifnot“/ImageSets”
247. ”/ImageSets”
248. ifnot“/ImageSets/Main”
249. ”/ImageSets/Main”
250. ”/wider_face_split/wider_face_”+img_set+“_bbx_gt.txt”
251. ”/ImageSets/Main/”+img_set+“.txt”,‘w’
252. ’r’
253. while(True
254. 1
255. if
256. break
257. ”/”,“_”
258. 4
259. ’\n’
261. forin
265. def
266. ”train”,“val”
267. forin
272. if“__main__”
274. ”/”+“train.txt”,rootdir+“/”+“trainval.txt”
275. ”/”+“val.txt”,rootdir+“/”+“test.txt”
276. ”/ImageSets/Main/”+“train.txt”,rootdir+“/ImageSets/Main/”+“trainval.txt”
277. ”/ImageSets/Main/”+“val.txt”,rootdir+“/ImageSets/Main/”+“test.txt”

import os,h5py,cv2,sys,shutil import numpy as np from xml.dom.minidom import Document rootdir="../" convet2yoloformat=True convert2vocformat=True resized_dim=(48, 48)      

文章来源: 将wierface标注转换为VOC格式