ansrb553326

(windows) jupyter notebook 실습환경

# ! conda install -c conda-forge xmltodict
# # xmltodict는 XML 파일을 parsing하여 Dictionary와 list 형태로 반환해줌.
import xmltodict

import os
import random

VOC_ROOT_DIR = "../../data/voc/VOCdevkit/VOC2012"
ANNO_DIR = os.path.join(VOC_ROOT_DIR, "Annotations")
JPEG_DIR = os.path.join(VOC_ROOT_DIR, "JPEGImages")

xml_files = os.listdir(ANNO_DIR)
print(xml_files[:5]); print(len(xml_files))

## xmltodict.parse()를 이용하여&nbsp; xml 파일을 Dictionary와 list형태로 변환
sample_xml = os.path.join(ANNO_DIR, '2007_000032.xml')
print('sample xml file : ', sample_xml)

with open(sample_xml, 'rb') as f: # notice the "rb" mode
&nbsp; &nbsp; parsed_dict = xmltodict.parse(f, xml_attribs = True)
&nbsp; &nbsp;&nbsp;
parsed_dict


#### Annotation 내의 Object들의 bounding box 정보를 이용하여 Bounding box 시각화

lmg_filename = parsed_dict['annotation']['filename']
lmg_filename = os.path.join(JPEG_DIR, lmg_filename)
print(lmg_filename)

lmg = cv2.imread(lmg_filename)
lmg_rgb = cv2.cvtColor(lmg, cv2.COLOR_BGR2RGB)

lmg_rgb_copy = img_rgb.copy()
green_rgb = (125,255,51)

for obj in parsed_dict['annotation']['object']:
&nbsp; &nbsp; class_name = obj['name']
&nbsp; &nbsp; bndbox = obj['bndbox']
&nbsp; &nbsp;&nbsp;
&nbsp; &nbsp; left = int(bndbox['xmin'])
&nbsp; &nbsp; top = int(bndbox['ymin'])
&nbsp; &nbsp; right = int(bndbox['xmax'])
&nbsp; &nbsp; bottom = int(bndbox['ymax'])
&nbsp; &nbsp;&nbsp;
&nbsp; &nbsp; cv2.rectangle(lmg_rgb_copy, (left, top), (right, bottom), color=green_rgb, thickness=2)
&nbsp; &nbsp;&nbsp;
plt.figure(figsize=(8,8))
plt.imshow(lmg_rgb_copy)
plt.show()

-------------



이 부분을 말씀하시는 것 같습니다.
7:00 ~

ansrb553326

Posts