def gen_xml(img_id):
img_desc = coco.loadImgs(img_id)[0]
fname = img_desc["file_name"]
# copy image
shutil.copy(f'{src_img_dir}/{fname}', f'{dst_img_dir}/{fname}')
# process annotations
bboxs = []
ann_ids = coco.getAnnIds(imgIds=img_desc['id'], catIds=cat_ids)
ann_descs = coco.loadAnns(ann_ids)
for ann_desc in ann_descs:
cat = COCO_CAT_NAMES[ann_desc['category_id']]
if cat in cat_names:
bbox = ann_desc['bbox']
xmin = int(bbox[0])
ymin = int(bbox[1])
xmax = int(bbox[2]) + int(bbox[0])
ymax = int(bbox[3]) + int(bbox[1])
bboxs.append((cat, xmin, ymin, xmax, ymax))
# get image info
img_path = f'{prefix_folder}{fname}'
img_w = img_desc["width"]
img_h = img_desc["height"]
img_d = 3
# save annotation info to xml file
dst_ann_path = f'{dst_ann_dir}/{fname[:-4]}.xml'
writer = pascal_voc_writer.Writer(img_path, img_w, img_h, img_d)
for bbox in bboxs:
writer.addObject(*bbox)
writer.save(dst_ann_path)
def gen_xml(ann_path):
fname = ann_path.stem
with open(ann_path, 'r') as f:
lines = f.readlines()
# calc bbox
bboxs = []
for line in lines:
obj = line.strip('\n').split(',')
if obj[5] in cat_dict:
xmin = int(obj[0])
ymin = int(obj[1])
xmax = int(obj[2]) + int(obj[0])
ymax = int(obj[3]) + int(obj[1])
bboxs.append((cat_dict[obj[5]], xmin, ymin, xmax, ymax))
# get image info
img = cv2.imread(f'{src_img_dir}/{fname}.jpg')
img_path = f'{prefix_folder}{fname}.jpg'
img_w = img.shape[1]
img_h = img.shape[0]
img_d = img.shape[2]
# save annotation info to xml file
dst_ann_path = f'{dst_ann_dir}/{fname}.xml'
writer = pascal_voc_writer.Writer(img_path, img_w, img_h, img_d)
for bbox in bboxs:
writer.addObject(*bbox)
writer.save(dst_ann_path)
def gen_xml(json_path):
fname = f'{json_path.parent.stem}/{json_path.name[:-21]}_leftImg8bit'
with open(json_path, 'r') as f:
data = json.load(f)
# calc bbox
bboxs = []
for obj in data["objects"]:
label = 'person' if obj["label"] == 'rider' else obj["label"]
polygon = obj["polygon"]
if label in cat_names:
x_coords, y_coords = zip(*polygon)
xmin = min(x_coords)
ymin = min(y_coords)
xmax = max(x_coords)
ymax = max(y_coords)
bboxs.append((label, xmin, ymin, xmax, ymax))
# get image info
img_path = f'{prefix_folder}{fname}.png'
img_w = data['imgWidth']
img_h = data['imgHeight']
img_d = 3
# save annotation info to xml file
dst_ann_path = f'{dst_ann_dir}/{fname}.xml'
writer = pascal_voc_writer.Writer(img_path, img_w, img_h, img_d)
for bbox in bboxs:
writer.addObject(*bbox)
writer.save(dst_ann_path)
def save_project_as_pascal_voc_detection(save_path, project: Project):
# Create root pascal 'datasets' folders
for dataset in project.datasets:
pascal_dataset_path = os.path.join(save_path, dataset.name)
pascal_dataset_relative_path = os.path.relpath(pascal_dataset_path,
save_path)
images_dir = os.path.join(pascal_dataset_path, 'JPEGImages')
anns_dir = os.path.join(pascal_dataset_path, 'Annotations')
lists_dir = os.path.join(pascal_dataset_path, 'ImageSets/Layout')
fs_utils.mkdir(pascal_dataset_path)
for subdir in [
'ImageSets', # Train list, Val list, etc.
'ImageSets/Layout',
'Annotations',
'JPEGImages'
]:
fs_utils.mkdir(os.path.join(pascal_dataset_path, subdir))
samples_by_tags = defaultdict(list) # TRAIN: [img_1, img2, ..]
for item_name in dataset:
img_path, ann_path = dataset.get_item_paths(item_name)
no_ext_name = fs_utils.get_file_name(item_name)
pascal_img_path = os.path.join(images_dir,
no_ext_name + OUT_IMG_EXT)
pascal_ann_path = os.path.join(anns_dir, no_ext_name + XML_EXT)
if item_name.endswith(OUT_IMG_EXT):
fs_utils.copy_file(img_path, pascal_img_path)
else:
img = image_utils.read(img_path)
image_utils.write(pascal_img_path, img)
ann = Annotation.load_json_file(ann_path,
project_meta=project.meta)
# Read tags for images lists generation
for tag in ann.img_tags:
samples_by_tags[tag.name].append(
(no_ext_name, len(ann.labels)))
writer = pascal_voc_writer.Writer(
path=pascal_dataset_relative_path,
width=ann.img_size[1],
height=ann.img_size[0])
for label in ann.labels:
obj_class = label.obj_class
rect: Rectangle = label.geometry.to_bbox()
writer.addObject(name=obj_class.name,
xmin=rect.left,
ymin=rect.top,
xmax=rect.right,
ymax=rect.bottom)
writer.save(pascal_ann_path)
save_images_lists(lists_dir, samples_by_tags)
def gen_xml(json_line):
data = eval(json_line)
# calc bboxs
bboxs = []
img_id = data['ID']
for gtbox in data['gtboxes']:
if gtbox['tag'] == 'person':
bbox = gtbox['vbox']
xmin = int(bbox[0])
ymin = int(bbox[1])
xmax = int(bbox[2]) + int(bbox[0])
ymax = int(bbox[3]) + int(bbox[1])
bboxs.append(('person', xmin, ymin, xmax, ymax))
# get image info
img = cv2.imread(f'{src_img_dir}/{img_id}.jpg')
img_path = f'{prefix_folder}{img_id}.jpg'
img_w = img.shape[1]
img_h = img.shape[0]
img_d = img.shape[2]
# save annotation info to xml file
dst_ann_path = f'{dst_ann_dir}/{img_id}.xml'
writer = pascal_voc_writer.Writer(img_path, img_w, img_h, img_d)
for bbox in bboxs:
writer.addObject(*bbox)
writer.save(dst_ann_path)
def process_detection_intermeddiate(self, frame, orig_box, image_path):
minx, miny, maxx, maxy, label, accuracy = orig_box
outputFrame = self.tf_detector.DisplayDetection(frame, orig_box)
if self.config.tf_od_frame_write:
cv2.imwrite(image_path, outputFrame)
if self.config.tf_od_annotation_write:
imH, imW, _ = frame.shape
writer = pascal_voc_writer.Writer(image_path, imW, imH)
minX, minY, maxX, maxY, klass, confidence = orig_box
writer.addObject(klass, minX, minY, maxX, maxY)
writer.save(image_path.replace('.jpg', '.xml'))
if self.config.show_fps:
cv2.putText(outputFrame, "%.2f fps" % self.fps.fps,
(10, outputFrame.shape[0] - 10),
cv2.FONT_HERSHEY_SIMPLEX, 0.35, (0, 255, 255), 1)
self.output_frame.enqueue(outputFrame)
def test(self):
writer = pascal_voc_writer.Writer(
path='hello.jpg',
width=600,
height=400,
)
writer.addObject(
name='category1',
xmin=10,
ymin=20,
xmax=30,
ymax=40,
)
writer.save('this.xml')
def set_writer(in_image_path, h, w, main_dir = default_dir):
global image_path, out_xml, writer
image_path = in_image_path
writer = pvw.Writer(image_path, h, w)
base_name = image_path.split('/')[-1]
out_xml = main_dir+'{}.xml'.format(base_name.split('.')[0])
im = Image.open('../dataset/train_images/' + your_list[i, 0] + '.jpg')
img_x, img_y = im.size
xmin = round(float(your_list[i, 4]) * img_x)
xmax = round(float(your_list[i, 5]) * img_x)
ymin = round(float(your_list[i, 6]) * img_y)
ymax = round(float(your_list[i, 7]) * img_y)
#
if xmax - xmin > img_x / 2:
print('$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$${}'.format(imgname))
if ymax - ymin > img_y / 2:
print('$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$${}'.format(imgname))
writer = pasc.Writer(your_list[i, 0] + '.jpg', img_x, img_y)
writer.addObject(in_person[classname], xmin, ymin, xmax, ymax)
print(good_img)
if good_img == 1:
if your_list[i + 1, 0] != imgname:
writer.save('xml_ann/' + imgname + '.xml')
# print(' saved : '+imgname+'.xml')
good_img = 0
if good_img == 1:
writer.save('xml_ann/' + imgname + '.xml')
#print(' saved last: '+imgname+'.xml')
