widerface数据库转voc2007数据集（python/matlab实现）
python实现基本需求，可以在此基础上修改
# -*- coding: utf-8 -*- """ Created on 17-5-27 @author: zly """ from skimage import io import shutil import random import os import string headstr = """\ <annotation> <folder>VOC2007</folder> <filename>%06d.jpg</filename> <source> <database>My Database</database> <annotation>PASCAL VOC2007</annotation> <image>flickr</image> <flickrid>NULL</flickrid> </source> <owner> <flickrid>NULL</flickrid> <name>company</name> </owner> <size> <width>%d</width> <height>%d</height> <depth>%d</depth> </size> <segmented>0</segmented> """ objstr = """\ <object> <name>%s</name> <pose>Unspecified</pose> <truncated>0</truncated> <difficult>0</difficult> <bndbox> <xmin>%d</xmin> <ymin>%d</ymin> <xmax>%d</xmax> <ymax>%d</ymax> </bndbox> </object> """ tailstr = '''\ </annotation> ''' def all_path(filename): return os.path.join('widerface', filename) def writexml(idx, head, bbxes, tail): filename = all_path("Annotations/%06d.xml" % (idx)) f = open(filename, "w") f.write(head) for bbx in bbxes: f.write(objstr % ('face', bbx[0], bbx[1], bbx[0] + bbx[2], bbx[1] + bbx[3])) f.write(tail) f.close() def clear_dir(): if shutil.os.path.exists(all_path('Annotations')): shutil.rmtree(all_path('Annotations')) if shutil.os.path.exists(all_path('ImageSets')): shutil.rmtree(all_path('ImageSets')) if shutil.os.path.exists(all_path('JPEGImages')): shutil.rmtree(all_path('JPEGImages')) shutil.os.mkdir(all_path('Annotations')) shutil.os.makedirs(all_path('ImageSets/Main')) shutil.os.mkdir(all_path('JPEGImages')) def excute_datasets(idx, datatype): f = open(all_path('ImageSets/Main/' + datatype + '.txt'), 'a') f_bbx = open(all_path('wider_face_split/wider_face_' + datatype + '_bbx_gt.txt'), 'r') while True: filename = string.strip(f_bbx.readline(), '\n') if not filename: break im = io.imread(all_path('WIDER_' + datatype + '/images/'+filename)) head = headstr % (idx, im.shape[1], im.shape[0], im.shape[2]) nums = string.strip(f_bbx.readline(), '\n') bbxes = [] for ind in xrange(string.atoi(nums)): bbx_info = string.split(string.strip(f_bbx.readline(), ' \n'), ' ') bbx = [string.atoi(bbx_info[i]) for i in range(len(bbx_info))] #x1, y1, w, h, blur, expression, illumination, invalid, occlusion, pose if bbx[7]==0: bbxes.append(bbx) writexml(idx, head, bbxes, tailstr) shutil.copyfile(all_path('WIDER_' + datatype + '/images/'+filename), all_path('JPEGImages/%06d.jpg' % (idx))) f.write('%06d\n' % (idx)) idx +=1 f.close() f_bbx.close() return idx # 打乱样本 def shuffle_file(filename): f = open(filename, 'r+') lines = f.readlines() random.shuffle(lines) f.seek(0) f.truncate() f.writelines(lines) f.close() if __name__ == '__main__': clear_dir() idx = 1 idx = excute_datasets(idx, 'train') idx = excute_datasets(idx, 'val')
目录结构如下
以下python代码为读取mat，废弃，仅供学习
import h5py
from skimage import io
import shutil
import random
headstr = """\
<annotation>
<folder>VOC2007</folder>
<filename>%06d.jpg</filename>
<source>
<database>My Database</database>
<annotation>PASCAL VOC2007</annotation>
<image>flickr</image>
<flickrid>NULL</flickrid>
</source>
<owner>
<flickrid>NULL</flickrid>
<name>facevise</name>
</owner>
<size>
<width>%d</width>
<height>%d</height>
<depth>%d</depth>
</size>
<segmented>0</segmented>
"""
objstr = """\
<object>
<name>%s</name>
<pose>Unspecified</pose>
<truncated>0</truncated>
<difficult>0</difficult>
<bndbox>
<xmin>%d</xmin>
<ymin>%d</ymin>
<xmax>%d</xmax>
<ymax>%d</ymax>
</bndbox>
</object>
"""
tailstr ='''\
</annotation>
'''
def writexml(idx, head, objs, tail):
filename = "Annotations/%06d.xml" % (idx)
f = open(filename, "w")
f.write(head)
f.write(objs)
f.write(tail)
f.close()
def clear_dir():
if shutil.os.path.exists('Annotations'):
shutil.rmtree('Annotations')
if shutil.os.path.exists('ImageSets'):
shutil.rmtree('ImageSets')
if shutil.os.path.exists('JPEGImages'):
shutil.rmtree('JPEGImages')
shutil.os.mkdir('Annotations')
shutil.os.makedirs('ImageSets/Main')
shutil.os.mkdir('JPEGImages')
def excute_datasets(idx, datatype):
f = open('ImageSets/Main/'+datatype+'.txt', 'a')
mat = h5py.File('wider_face_split/wider_face_'+datatype+'.mat', 'r')
file_list = mat['file_list'][:]
event_list = mat['event_list'][:]
bbx_list = mat['face_bbx_list'][:]
for i in range(file_list.size):
file_list_sub = mat[file_list[0,i]][:]
bbx_list_sub = mat[bbx_list[0, i]][:]
event_value = ''.join(chr(x) for x in mat[event_list[0,i]][:])
for j in range(file_list_sub.size):
root = 'WIDER_'+datatype+'/images/'+event_value+'/'
filename = root + ''.join([chr(x) for x in mat[file_list_sub[0, j]][:]])+'.jpg'
im = io.imread(filename)
head = headstr % (idx, im.shape[1], im.shape[0], im.shape[2])
bboxes = mat[bbx_list_sub[0, j]][:]
objs = ''.join([objstr % ('face', \
bboxes[0,k],bboxes[1,k], bboxes[0,k]+bboxes[2,k]-1,bboxes[1,k]+bboxes[3,k]-1) \
for k in range(bboxes.shape[1])])
writexml(idx, head, objs, tailstr)
shutil.copyfile(filename, 'JPEGImages/%06d.jpg' % (idx))
f.write('%06d\n' % (idx))
idx +=1
f.close()
return idx
#打乱样本
def shuffle_file(filename):
f = open(filename, 'r+')
lines = f.readlines()
random.shuffle(lines)
f.seek(0)
f.truncate()
f.writelines(lines)
f.close()
if __name__ == '__main__':
clear_dir()
idx = 1
idx = excute_datasets(idx, 'train')
idx = excute_datasets(idx, 'val')
matlab实现
head.xml
<annotation> <folder>widerface</folder> <filename>%06d.jpg</filename> <source> <database>My Database</database> <annotation>VOC2007</annotation> <image>flickr</image> <flickrid>NULL</flickrid> </source> <owner> <flickrid>NULL</flickrid> <name>facevise</name> </owner> <size> <width>%d</width> <height>%d</height> <depth>%d</depth> </size> <segmented>0</segmented>
object.xml
<object> <name>%s</name> <pose>Unspecified</pose> <truncated>0</truncated> <difficult>0</difficult> <bndbox> <xmin>%d</xmin> <ymin>%d</ymin> <xmax>%d</xmax> <ymax>%d</ymax> </bndbox> </object>
tail.xml
</annotation>
function WiderFace2VOC() %% wider face % The corresponding annotations are in the following format: % Here, each face bounding boxe is denoted by: % <x_left y_top width height>. %% voc % 000001.jpg car 44 28 132 121 %前面是图片名，中间是目标类别，最后是目标的包围框坐标（左上角和右下角坐标）。 %% clc; clear; fclose all; [~, ~, ~] = rmdir('Annotations', 's'); [~, ~, ~] = rmdir('ImageSets', 's'); [~, ~, ~] = rmdir('JPEGImages', 's'); [~, ~, ~] = mkdir('Annotations'); [~, ~, ~] = mkdir('ImageSets/Main'); [~, ~, ~] = mkdir('JPEGImages'); train_root = 'WIDER_train/images'; split_file = 'wider_face_split/wider_face_train'; data = load(split_file); headXml = fopen('head.xml', 'r'); headXmlFormat = fread(headXml, Inf, '*char'); fclose(headXml); objectXml = fopen('object.xml', 'r'); objectXmlFormat = fread(objectXml, Inf, '*char'); fclose(objectXml); tailXml = fopen('tail.xml', 'r'); tailXmlFormat = fread(tailXml, Inf, '*char'); fclose(tailXml); trainID = fopen('ImageSets/Main/train.txt', 'w'); trainvalID = fopen('ImageSets/Main/trainval.txt', 'w'); valID = fopen('ImageSets/Main/val.txt', 'w'); testID = fopen('ImageSets/Main/test.txt', 'w'); idx = 1; for i=1:numel(data.event_list) for j=1:numel(data.file_list{i}) imagename = fullfile(train_root, data.event_list{i}, strcat(data.file_list{i}{j}, '.jpg')); sz = size(imread(imagename)); AnnotationsXml = fopen(sprintf('Annotations/%06d.xml', idx), 'w'); fprintf(AnnotationsXml, headXmlFormat, idx, sz(2), sz(1),sz(3)); for k = 1:size(data.face_bbx_list{i}{j}, 1) rc = data.face_bbx_list{i}{j}(k, :); rc = round([rc(1), rc(2), rc(1)+rc(3)-1, rc(2)+rc(4)-1]); fprintf(AnnotationsXml, objectXmlFormat, 'face', rc(1), rc(2), rc(3), rc(4)); end fprintf(AnnotationsXml, tailXmlFormat); fprintf(trainID, '%06d\n', idx); fprintf(trainvalID, '%06d\n', idx); fclose(AnnotationsXml); copyfile(imagename, sprintf('JPEGImages/%06d.jpg', idx)); idx = idx + 1; end disp(i); end train_root = 'WIDER_val/images'; split_file = 'wider_face_split/wider_face_val'; data = load(split_file); for i=1:numel(data.event_list) for j=1:numel(data.file_list{i}) imagename = fullfile(train_root, data.event_list{i}, strcat(data.file_list{i}{j}, '.jpg')); sz = size(imread(imagename)); AnnotationsXml = fopen(sprintf('Annotations/%06d.xml', idx), 'w'); fprintf(AnnotationsXml, headXmlFormat, idx, sz(2), sz(1),sz(3)); for k = 1:size(data.face_bbx_list{i}{j}, 1) rc = data.face_bbx_list{i}{j}(k, :); rc = round([rc(1), rc(2), rc(1)+rc(3)-1, rc(2)+rc(4)-1]); fprintf(AnnotationsXml, objectXmlFormat, 'face', rc(1), rc(2), rc(3), rc(4)); end fprintf(AnnotationsXml, tailXmlFormat); if mod(idx, 2) fprintf(valID, '%06d\n', idx); fprintf(trainvalID, '%06d\n', idx); else fprintf(testID, '%06d\n', idx); end fclose(AnnotationsXml); copyfile(imagename, sprintf('JPEGImages/%06d.jpg', idx)); idx = idx+1; end disp(i); end fclose(trainID); fclose(trainvalID); fclose(valID); fclose(testID); fclose all;
