def __init__(self, opt):
BaseDataset.__init__(self, opt)
self.image_list = util.get_file_list(
os.path.join(self.opt.data_dir, 'crop'))
self.landmark_dict = self.load_landmark_dict()
self.transforms_input = transforms.Compose([
transforms.Resize((224, 224)),
transforms.ToTensor(),
transforms.Normalize(mean=[0.5141, 0.4074, 0.3588],
std=[1.0, 1.0, 1.0])
])
self.transforms_gt = transforms.ToTensor()
def __init__(self, opt):
self.opt = opt
self.data_dir = opt.data_dir
self.Nw = opt.Nw
self.feature_list = util.load_coef(
os.path.join(self.data_dir, 'feature'))
self.filenames = util.get_file_list(
os.path.join(self.data_dir, 'feature'))
if opt.isTrain:
self.alpha_list = util.load_coef(
os.path.join(self.data_dir, 'alpha'))
self.beta_list = util.load_coef(os.path.join(
self.data_dir, 'beta'))
self.delta_list = util.load_coef(
os.path.join(self.data_dir, 'delta'))
self.gamma_list = util.load_coef(
os.path.join(self.data_dir, 'gamma'))
self.rotation_list = util.load_coef(
os.path.join(self.data_dir, 'rotation'))
self.translation_list = util.load_coef(
os.path.join(self.data_dir, 'translation'))
def init_valid_dataset(config):
return create_dataset(
get_file_list(config.get("data", "valid_data_path"),
config.get("data", "valid_file_list")), config,
config.getint("reader", "valid_reader_num"), "valid")
def init_train_dataset(config):
return create_dataset(
get_file_list(config.get("data", "train_data_path"),
config.get("data", "train_file_list")), config,
config.getint("reader", "train_reader_num"), "train")
if __name__ == '__main__':
import pathlib
from tqdm import tqdm
import matplotlib.pyplot as plt
from utils.util import show_img, draw_bbox, save_result, get_file_list, crop_bbox
args = init_args()
print(args)
# 初始化网络
model = DBNet(args.model_path,
post_p_thre=args.thre,
gpu_id=0,
imageH=960,
imageW=480)
img_folder = pathlib.Path(args.input_folder)
for img_path in tqdm(get_file_list(args.input_folder, p_postfix=['.jpg'])):
img = cv2.imread(img_path)
# if img.shape[0] /img.shape[1] > 2 or img.shape[1]/img.shape[0] > 2:
# continue
preds, boxes_list, score_list, t = model.predict(
img_path, is_output_polygon=args.polygon, runtime='trt')
print('time cost: {}s'.format(t))
crops = crop_bbox(img[:, :, ::-1], boxes_list)
img = draw_bbox(img[:, :, ::-1], boxes_list)
if args.show:
show_img(preds)
show_img(img, title=os.path.basename(img_path))
plt.show()
# 保存结果到路径
os.makedirs(args.output_folder, exist_ok=True)
img_path = pathlib.Path(img_path)
# def
if __name__ == '__main__':
import pathlib
from tqdm import tqdm
import matplotlib.pyplot as plt
from utils.util import show_img, draw_bbox, save_result, get_file_list
args = init_args()
print(args)
os.environ['CUDA_VISIBLE_DEVICES'] = str('0')
# 初始化网络 0.1
model = Pytorch_model(args.model_path, post_p_thre=args.thre, gpu_id=0)
img_folder = pathlib.Path(args.input_folder) # dbnet/test/input/
for img_path in tqdm(
get_file_list(args.input_folder, p_postfix=['.jpg', '.png'])
): # img_path /home/share/gaoluoluo/dbnet/test/input/2018实验仪器发票.jpg
# print("img_path:",img_path) /home/share/gaoluoluo/dbnet/test/input/2018实验仪器发票.jpg
preds, boxes_list, score_list, t = model.predict(
img_path, is_output_polygon=args.polygon)
# print("preds:",preds)
print("boxes_list.shape:", np.shape(boxes_list))
print("boxes_list:", boxes_list) # 4 个点的坐标
# print("score_list:",score_list) # 率
print("t:", t) # 用时
# box = np.array()
box = []
print(len(boxes_list))
print(len(boxes_list[0]))
print(len(boxes_list[0][0]))
for i in range(0, len(boxes_list)):
import os
import cv2
import numpy as np
from tqdm import tqdm
from options.options import Options
from utils.util import create_dir, get_file_list
if __name__ == '__main__':
opt = Options().parse_args()
create_dir(os.path.join(opt.src_dir, 'comp'))
foregrounds = get_file_list(os.path.join(opt.src_dir, 'images'), suffix='fake')
backgrounds = get_file_list(os.path.join(opt.tgt_dir, 'crop'))
masks = get_file_list(os.path.join(opt.tgt_dir, 'mask'))
for i in tqdm(range(len(foregrounds))):
fg = cv2.imread(foregrounds[i])
bg = cv2.imread(backgrounds[i + opt.offset])
mask = cv2.imread(masks[i + opt.offset])
mask = cv2.erode(mask, np.ones((3,3), np.uint8), iterations=9)
mask = cv2.GaussianBlur(mask, (5,5), cv2.BORDER_DEFAULT) / 255.0
comp = mask * fg + (1 - mask) * bg
cv2.imwrite(os.path.join(opt.src_dir, 'comp', '%05d.png' % (i+1)), comp)
if i >= opt.test_num:
return args
if __name__ == '__main__':
import pathlib
from tqdm import tqdm
import matplotlib.pyplot as plt
from utils.util import show_img, draw_bbox, save_result, get_file_list
args = init_args()
print(args)
os.environ['CUDA_VISIBLE_DEVICES'] = str('0')
# 初始化网络 0.1
model = Pytorch_model(args.model_path, post_p_thre=args.thre, gpu_id=0)
img_folder = pathlib.Path(args.input_folder)# dbnet/test/input/
for img_path in tqdm(get_file_list(args.input_folder, p_postfix=['.jpg', '.png'])): # img_path /home/share/gaoluoluo/dbnet/test/input/2018实验仪器发票.jpg
# print("img_path:",img_path) /home/share/gaoluoluo/dbnet/test/input/2018实验仪器发票.jpg
preds, boxes_list, score_list, t = model.predict(img_path, is_output_polygon=args.polygon)
print("preds:",preds)
print("boxes_list.shape:",np.shape(boxes_list))
print("boxes_list:",boxes_list) # 4 个点的坐标
print("score_list:",score_list) # 率
print("t:",t) # 用时
# box = np.array()
box = []
for i in range(len(boxes_list)-1):
for j in range(len(boxes_list[0])-1):
box.append(np.array(boxes_list[i][j]))
print("box.shape:",np.shape(box))
print("np.array(box):",np.array(box))
