def __init__(self, cfg):
img_path = os.path.join(cfg.log_path, cfg.exp_name, 'train_images_*')
if glob.glob(img_path):
raise Exception('all directories with name train_images_* under '
'the experiment directory need to be removed')
path = os.path.join(cfg.log_path, cfg.exp_name)
self.model = MODELS[cfg.gan_type](cfg)
self.model.save_model(path, 0, 0)
if cfg.load_snapshot is not None:
self.model.load_model(cfg.load_snapshot)
shuffle = cfg.gan_type != 'recurrent_gan'
self.dataset = DATASETS[cfg.dataset](path=keys[cfg.dataset],
cfg=cfg,
img_size=cfg.img_size)
self.dataloader = DataLoader(self.dataset,
batch_size=cfg.batch_size,
shuffle=shuffle,
num_workers=cfg.num_workers,
pin_memory=True,
drop_last=True)
if cfg.dataset == 'codraw':
self.dataloader.collate_fn = codraw_dataset.collate_data
elif cfg.dataset == 'iclevr':
self.dataloader.collate_fn = clevr_dataset.collate_data
self.visualizer = VisdomPlotter(env_name=cfg.exp_name,
server=cfg.vis_server)
self.logger = None
self.cfg = cfg
def __init__(self, cfg):
img_path = os.path.join(cfg.log_path, cfg.exp_name, 'train_images_*')
if glob.glob(img_path):
raise Exception('all directories with name train_images_* under '
'the experiment directory need to be removed')
path = os.path.join(cfg.log_path, cfg.exp_name)
shuffle = False
print(keys[cfg.dataset])
self.dataset = DATASETS[cfg.dataset](path=keys[cfg.dataset],
cfg=cfg,
img_size=cfg.img_size)
# update the cfg's vocab_size
cfg.vocab_size = self.dataset.vocab_size
self.dataloader = DataLoader(self.dataset,
batch_size=cfg.batch_size,
shuffle=shuffle,
num_workers=cfg.num_workers,
pin_memory=True,
drop_last=True)
if cfg.dataset in ['codraw', 'codrawDialog']:
self.dataloader.collate_fn = codraw_dataset.collate_data
elif cfg.dataset == 'iclevr':
self.dataloader.collate_fn = clevr_dataset.collate_data
elif cfg.dataset in [
"gandraw", "gandraw_clean", "gandraw_64", "gandraw_64_DA"
]:
self.dataloader.collate_fn = gandraw_dataset.collate_data
self.model = MODELS[cfg.gan_type](cfg)
# Added by Mingyang
# if cfg.gan_type == "recurrent_gan_mingyang":
# print("Wrap DataParallel Around the whole model")
# self.model = DataParallel(self.model)
self.model.save_model(path, 0, 0)
if cfg.load_snapshot is not None:
print("load the model from: {}".format(cfg.load_snapshot))
self.model.load_model(cfg.load_snapshot)
self.visualizer = VisdomPlotter(env_name=cfg.exp_name,
server=cfg.vis_server)
self.logger = None
self.cfg = cfg
def __init__(self, cfg):
img_path = os.path.join(cfg.log_path,
cfg.exp_name,
'train_images_*')
if glob.glob(img_path):
raise Exception('all directories with name train_images_* under '
'the experiment directory need to be removed')
path = os.path.join(cfg.log_path, cfg.exp_name)
self.dataset = DATASETS[cfg.dataset](
path=keys[cfg.dataset], cfg=cfg, img_size=cfg.img_size)
# update the cfg's vocab_size
cfg.vocab_size = self.dataset.vocab_size
self.dataloader = DataLoader(self.dataset,
batch_size=cfg.batch_size,
shuffle=False,
num_workers=cfg.num_workers,
pin_memory=True,
drop_last=True)
if cfg.dataset in ['codraw', 'codrawDialog']:
self.dataloader.collate_fn = codraw_dataset.collate_data
elif cfg.dataset in ["gandraw", "gandraw_clean", "gandraw_64", "gandraw_64_DA"]:
self.dataloader.collate_fn = gandraw_dataset.collate_data
####################Load the Model###################
assert cfg.gan_type == "recurrent_gan_teller", "To run a teller trainer, you will need to use 'recurrent_gan_teller' as gan_type"
self.model = MODELS[cfg.gan_type](cfg)
self.model.save_model(path, 0, 0)
#####################################################
# Launch the visualizer
self.visualizer = VisdomPlotter(
env_name=cfg.exp_name, server=cfg.vis_server)
self.logger = None
self.cfg = cfg
from torch import nn
import torch.backends.cudnn as cudnn
import torchvision.transforms as transforms
from tqdm import tqdm
from geneva.models.object_localizer import inception_v3
from geneva.utils.config import keys
from geneva.utils.visualize import VisdomPlotter
model_urls = {
# Inception v3 ported from TensorFlow
'inception_v3_google': 'https://download.pytorch.org/models/inception_v3_google-1a9a5a14.pth',
}
viz = VisdomPlotter(env_name='inception_localizer', server='http://localhost')
class Inception3ObjectLocalizer(nn.Module):
def __init__(self, num_objects=58, pretrained=True, num_coords=3):
super().__init__()
self.inception3 = inception_v3(pretrained=pretrained)
self.inception3.fc = nn.Linear(self.inception3.fc.in_features, 512)
self.detector = nn.Sequential(nn.Linear(512, 256),
nn.Linear(256, num_objects),
nn.Sigmoid())
self.localizer = nn.Sequential(nn.Linear(1024, 512),
nn.Linear(512, num_objects * num_coords))
self.num_objects = num_objects
self.num_coords = num_coords
del tester
torch.cuda.empty_cache()
metrics_report = dict()
if self.cfg.metric_inception_objects:
io_jss, io_ap, io_ar, io_af1, io_cs, io_gs = \
report_inception_objects_score(self.visualizer,
self.logger,
iteration,
self.cfg.results_path,
keys[self.cfg.dataset + '_inception_objects'],
keys[self.cfg.val_dataset],
self.cfg.dataset)
metrics_report['jaccard'] = io_jss
metrics_report['precision'] = io_ap
metrics_report['recall'] = io_ar
metrics_report['f1'] = io_af1
metrics_report['cossim'] = io_cs
metrics_report['relsim'] = io_gs
return metrics_report
if __name__ == '__main__':
cfg = parse_config()
visualizer = VisdomPlotter(env_name=cfg.exp_name)
logger = None
dataset = cfg.dataset
evaluator = Evaluator(cfg, visualizer, logger, dataset)
evaluator.evaluate()
from geneva.data.datasets import DATASETS
from geneva.evaluation.evaluate import Evaluator
from geneva.utils.config import keys, parse_config
from geneva.utils.visualize import VisdomPlotter
from geneva.models.models import MODELS
from geneva.data import gandraw_dataset
import time
if __name__ == '__main__':
config_file = "example_args/gandraw_args.json"
with open(config_file, 'r') as f:
cfg = json.load(f)
cfg = easydict.EasyDict(cfg)
best_iteration = 1500 #Manually define
#Initialize the evaluator
visualizer = VisdomPlotter(env_name=cfg.exp_name, server=cfg.vis_server)
logger = None
evaluator = Evaluator.factory(cfg, visualizer, logger, visualize_images=[])
metric_report = evaluator.evaluate(best_iteration, use_test=True)
print("evaluation results for iter: {} on test data: \n".format(
best_iteration))
for key, value in metric_report.items():
print("{metric_name}: {metric_value}; \n".format(metric_name=key,
metric_value=value))