将SYSU数据集转换为VOC--只关注行人检测

 1. 构建VOC文件夹

import os

def make_voc_dir():
    # labels 目录若不存在，创建labels目录。若存在，则清空目录
    if not os.path.exists('../VOC2007/Annotations'):
        os.makedirs('../VOC2007/Annotations')
    if not os.path.exists('../VOC2007/ImageSets'):
        os.makedirs('../VOC2007/ImageSets')
        os.makedirs('../VOC2007/ImageSets/Main')
    if not os.path.exists('../VOC2007/JPEGImages'):
        os.makedirs('../VOC2007/JPEGImages')

if __name__ == '__main__':
    make_voc_dir()

 2. 将SYSU数据集的所有图片复制到VOC2007/JPEGImages

3. 生成XML标注文件

import os
import numpy as np
import scipy.io as sio
import shutil
from lxml.etree import Element, SubElement, tostring
from xml.dom.minidom import parseString

def process_annotations(root_dir, VOCRoot):

    annotation_dir = os.path.join(root_dir, 'annotation')
    imgs = sio.loadmat(os.path.join(
        annotation_dir, 'Images.mat'))['Img'].squeeze()

    imnames = []
    imnames_unique = []
    all_bboxes = []

    for im_name, _, boxes in imgs:
        filename = str(im_name[0]) # 's14859.jpg'
        bboxes = np.asarray([b[0][0] for b in boxes[0]]) # <class 'tuple'>: (9, 4)
        valid_index = np.where((bboxes[:, 2] > 0) & (bboxes[:, 3] > 0))[0]
        assert valid_index.size > 0, \
            'Warning: {} has no valid boxes.'.format(filename)
        bboxes = bboxes[valid_index].astype(np.int32)

        node_root = Element('annotation')
        node_folder = SubElement(node_root, 'folder')
        node_folder.text = 'JPEGImages'
        node_filename = SubElement(node_root, 'filename')
        node_filename.text = 'VOC2007/JPEGImages/%s' % filename
        node_size = SubElement(node_root, 'size')
        node_width = SubElement(node_size, 'width')
        node_width.text = '%s' % 640
        node_height = SubElement(node_size, 'height')
        node_height.text = '%s' % 480
        node_depth = SubElement(node_size, 'depth')
        node_depth.text = '3'

        for box_idx, box in enumerate(boxes.T):
            box = box[0][0][0]
            xmin = box[0] + 1
            ymin = box[1] + 1
            obj_width= box[2]
            obj_height = box[3]
            xmax = xmin + obj_width
            ymax = ymin + obj_height
            difficult = 0

            if obj_height <= 4 or obj_width <= 4:
                difficult = 1

            node_object = SubElement(node_root, 'object')
            node_name = SubElement(node_object, 'name')
            node_name.text = 'person'
            node_difficult = SubElement(node_object, 'difficult')
            node_difficult.text = '%s' % difficult
            node_bndbox = SubElement(node_object, 'bndbox')
            node_xmin = SubElement(node_bndbox, 'xmin')
            node_xmin.text = '%s' % xmin
            node_ymin = SubElement(node_bndbox, 'ymin')
            node_ymin.text = '%s' % ymin
            node_xmax = SubElement(node_bndbox, 'xmax')
            node_xmax.text = '%s' % xmax
            node_ymax = SubElement(node_bndbox, 'ymax')
            node_ymax.text = '%s' % ymax
            node_name = SubElement(node_object, 'pose')
            node_name.text = 'Unspecified'
            node_name = SubElement(node_object, 'truncated')
            node_name.text = '0'

        image_path = VOCRoot + '/JPEGImages/' + filename
        xml = tostring(node_root, pretty_print=True)  # 'annotation'
        dom = parseString(xml)
        xml_name = filename.replace('.jpg', '.xml')
        xml_path = VOCRoot + '/Annotations/' + xml_name
        with open(xml_path, 'wb') as f:
            f.write(xml)


if __name__ == '__main__':
    root_dir = '../sysu'
    VOCRoot = '../VOC2007'
    print('Processing the mat files...')
    process_annotations(root_dir, VOCRoot)