pascal voc target detection data set in the following format:
among them:
- Annotations for the image annotation information xml file
- ImageSets training set, test set, verification, validation set of training images txt file name
- JPEGImages to the original picture
Pascal voc or yolo data format may be used for labeling labelimg: Download:
Link: https: //pan.baidu.com/s/1r8x7tu0sdO_UUuCXKVfELQ
extraction code: l325
Very simple operation, not introduced.
Marked good xml file similar to the following:
<annotation> <folder>JPEGImages</folder> <filename>test_00000002.jpg</filename> <path>E:\detection\pascal voc\maskornot\JPEGImages\test_00000002.jpg</path> <source> <database>Unknown</database> </source> <size> <width>480</width> <height>600</height> <depth>3</depth> </size> <segmented>0</segmented> <object> <name>mask</name> <pose>Unspecified</pose> <truncated>0</truncated> <difficult>0</difficult> <bndbox> <xmin>112</xmin> <ymin>7</ymin> <xmax>352</xmax> <ymax>325</ymax> </bndbox> </object> </annotation>
An image corresponding follows:
Then divide the training set, test set, validation set, validation set of training: focused on the original VOC2007 data, trainval accounts for about 50% of the entire data set, test about 50% of the entire data set; train of about 50% trainval, val of about 50% trainval
import os import random trainval_percent = 0.5 train_percent = 0.5 xmlfilepath = '/content/drive/My Drive/pytorch_ssd/data/maskornot/Annotations' txtsavepath = '/content/drive/My Drive/pytorch_ssd/data/maskornot/ImageSets/Main' total_xml = os.listdir(xmlfilepath) num=len(total_xml) list=range(num) tv=int(num*trainval_percent) tr=int(tv*train_percent) trainval= random.sample(list,tv) train=random.sample(trainval,tr) ftrainval = open(txtsavepath+'/trainval.txt', 'w') ftest = open(txtsavepath+'/test.txt', 'w') ftrain = open(txtsavepath+'/train.txt', 'w') fval = open(txtsavepath+'/val.txt', 'w') for i in list: name=total_xml[i][:-4]+'\n' if i in trainval: ftrainval.write(name) if i in train: ftrain.write(name) else: fval.write(name) else: ftest.write(name) ftrainval.close() ftrain.close() fval.close() ftest.close()
After running:
Wherein tranval.txt partial result is:
test_00000002
test_00000003
test_00000006
test_00000009
test_00000008
test_00000012
test_00000013
test_00000014
test_00000020
At this point, create target detection data set is complete.
The next section, using pytorch-ssd training to create their own data sets.