# This Python program is based on features added in Python 3.5. Python 3.7 is recommanded.
# Requires Pillow >= 6.0.0 for ImageOps.exif_transpose() in utils.py
from PIL import Image
import math
from os import walk, listdir
from os.path import isfile, join
import config
from image import is_animated, rotate, combineImages
from video import animate, getFps, avgFps
# Load configuration
args = config.args()
userconf = config.user(args.config)
CONFIG = config.general(args, userconf)
CANVAS_ATT = config.canvas(args, userconf)
FFMPEG = config.ffmpeg(args, userconf)
# Walk in subdirectories (jour 1, jour 2, ...)
for root, dirs, _ in walk(CONFIG["ROOT_DIR"]):
print("dirs in root: " + str(dirs)) # Debug
# For each subdir
for subdir in dirs:
dir = join(root, subdir)
print("current subdir: " + str(dir)) # Debug
# Read all the files in the dir & keep only images
images = []
for file in listdir(dir):
file = join(dir, file)
try:
import torch
import torchvision
from torch.utils.data import DataLoader
import torch.nn.functional as F
import config
from torchvision import transforms
from model import SiameseNetwork
from utils import SiameseNetworkDataset, imshow
opt = config.args()
folder_dataset_test = torchvision.datasets.ImageFolder(root=opt.testing_dir)
# 定义图像dataset
transform_test = transforms.Compose([transforms.Resize((100, 100)),
transforms.ToTensor()])
siamese_dataset_test = SiameseNetworkDataset(imageFolderDataset=folder_dataset_test,
transform=transform_test,
should_invert=False)
# 定义图像dataloader
test_dataloader = DataLoader(siamese_dataset_test,
shuffle=True,
batch_size=1)
net = SiameseNetwork().cuda()
net.load_state_dict(torch.load('checkpoint/siameseNet49.pth'))
# 生成对比图像
dataiter = iter(test_dataloader)
x0, _, _ = next(dataiter)
for i in range(10):
if best_eval['loss'] == 0 or f1 > best_eval['f1']:
best_eval['loss'] = epoch_loss
best_eval['acc'] = acc
best_eval['precision'] = precision
best_eval['recall'] = recall
best_eval['f1'] = f1
torch.save(bert_crf_tagger,
f'{paras.log_save_path}/{paras.model_save_name}')
with open(f'{paras.log_save_path}/checkpoint.log', 'w') as wf:
wf.write(
f'Save time: {time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())}\n'
)
wf.write(f'Best F1-score: {best_eval["f1"]:.4f}\n')
wf.write(f'Best Precision: {best_eval["precision"]:.4f}\n')
wf.write(f'Best Recall: {best_eval["recall"]:.4f}\n')
wf.write(f'Best Accuracy: {best_eval["acc"]:.4f}\n')
wf.write(f'Best Loss: {best_eval["loss"]:.4f}\n')
logger.info(
f'Updated model, best F1-score: {best_eval["f1"]:.4f}\n')
if __name__ == '__main__':
args = args()
set_seed(args.seed)
main(args)