def __init__(self):
self.VocTrainDataLoader = DataLoader(COCODataset(is_train=True),
batch_size=opt.batch_size,
shuffle=True,
num_workers=opt.num_workers,
drop_last=True,
pin_memory=True)
self.VocTestDataLoader = DataLoader(COCODataset(is_train=False),
batch_size=1,
shuffle=False,
num_workers=0,
drop_last=True,
pin_memory=True)
self.img_size = opt.img_size
self.anchors = parse_anchors(opt.anchors_path)
self.train_data_length = len(COCODataset(is_train=True))
self.val_data_length = len(COCODataset(is_train=False))
self.trainer = Yolov3COCOTrainer()
self.testDataset = COCODataset(is_train=False)
self.test_imgs = np.random.randint(0, self.val_data_length, 10)
# self.test_imgs = np.random.randint(low=0, high=len(self.testDataset), size=35)
self.normailze = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize(opt.mean, opt.std)])
self.logger = opt.logger
self.writer = SummaryWriter(log_dir=os.path.join(
opt.base_path, 'logs', f'summary_{opt.optimizer}'))
def main():
print("ann file:", annotations_file_path)
print("root:", sperm_images_path)
albumentations_transforms = get_albumentations_transforms( mode = 'train')
dataset = COCODataset(root_dir = sperm_images_path, coco_path=annotations_file_path, transforms = albumentations_transforms)
image, target = dataset[0]
print(type(image))
print(type(target))
print(target.keys())
def test(args):
dataset_dir = args.dataset_dir
with open(args.config_file, 'r') as f:
try:
config = yaml.safe_load(f)
except yaml.YAMLError as exc:
logging.error("Invalid YAML")
logging.error(exc)
if (args.dataset_type == 0):
coco_dataset_dir = path.join(dataset_dir, 'coco')
train_dataset = COCODataset(coco_dataset_dir, config, mode='train')
val_dataset = COCODataset(coco_dataset_dir, config, mode='validation')
else:
raise NotImplementedError
images, corrs = train_dataset[0]
print(images.shape, corrs.shape)
print('training image loading working properly')
images, corrs = val_dataset[0]
print(images.shape, corrs.shape)
print('validation image loading working properly')
def test_COCODataset(self):
from transform import get_transforms
from dataset import COCODataset
from utils import read_yaml
# read config file for transforms
config_path = "configs/default_config.yml"
config = read_yaml(config_path)
# form basic albumentation transform
transforms = get_transforms(config=config, mode="val")
# init COCODataset
DATA_ROOT = "tests/data/"
COCO_PATH = "tests/data/coco_true_1.json"
dataset = COCODataset(DATA_ROOT, COCO_PATH, transforms)
# iterate over the dataset
# and get annotations (target dict) for the first image
image_tensor = next(iter(dataset))[0]
target_tensor = next(iter(dataset))[1]
# apply checks for image tensor
self.assertEqual(image_tensor.type(), "torch.FloatTensor")
self.assertEqual(list(image_tensor.size()), [3, 1920, 1080])
self.assertAlmostEqual(float(image_tensor.max()), 1.0, places=2)
self.assertAlmostEqual(float(image_tensor.mean()), 0.39, places=2)
# apply checks for each field in the target tensor dict
boxes_tensor_0 = target_tensor["boxes"][0]
self.assertEqual(boxes_tensor_0.type(), "torch.FloatTensor")
self.assertEqual(boxes_tensor_0.cpu().numpy().tolist(),
[97.0, 643.0, 931.0, 1185.0])
labels_tensor_0 = target_tensor["labels"][0]
self.assertEqual(labels_tensor_0.type(), "torch.LongTensor")
self.assertEqual(labels_tensor_0.cpu().numpy().item(), 1)
masks_tensor_0 = target_tensor["masks"][0]
self.assertEqual(masks_tensor_0.type(), "torch.ByteTensor")
self.assertEqual(list(masks_tensor_0.size()), [1920, 1080])
self.assertAlmostEqual(float(masks_tensor_0.max()), 1.0, places=1)
image_id_tensor_0 = target_tensor["image_id"][0]
self.assertEqual(image_id_tensor_0.type(), "torch.LongTensor")
self.assertEqual(image_id_tensor_0.cpu().numpy().item(), 0)
area_tensor_0 = target_tensor["area"][0]
self.assertEqual(area_tensor_0.type(), "torch.FloatTensor")
self.assertEqual(area_tensor_0.cpu().numpy().item(), 452028.0)
iscrowd_tensor_0 = target_tensor["iscrowd"][0]
self.assertEqual(iscrowd_tensor_0.type(), "torch.LongTensor")
self.assertEqual(iscrowd_tensor_0.cpu().numpy().item(), 0)
boxes_tensor_1 = target_tensor["boxes"][1]
self.assertEqual(boxes_tensor_1.type(), "torch.FloatTensor")
self.assertEqual(boxes_tensor_1.cpu().numpy().tolist(),
[97.0, 500.0, 931.0, 1185.0])
labels_tensor_1 = target_tensor["labels"][1]
self.assertEqual(labels_tensor_1.type(), "torch.LongTensor")
self.assertEqual(labels_tensor_1.cpu().numpy().item(), 2)
masks_tensor_1 = target_tensor["masks"][1]
self.assertEqual(masks_tensor_1.type(), "torch.ByteTensor")
self.assertEqual(list(masks_tensor_1.size()), [1920, 1080])
self.assertAlmostEqual(float(masks_tensor_1.max()), 1.0, places=1)
area_tensor_1 = target_tensor["area"][1]
self.assertEqual(area_tensor_1.type(), "torch.FloatTensor")
self.assertEqual(area_tensor_1.cpu().numpy().item(), 571290.0)
os.makedirs(working_dir)
logger = SummaryWriter(working_dir)
n_gpu = int(os.environ['WORLD_SIZE']) if 'WORLD_SIZE' in os.environ else 1
args.n_gpu = n_gpu
args.distributed = n_gpu > 1
if args.distributed:
torch.cuda.set_device(args.local_rank)
torch.distributed.init_process_group(backend='gloo',
init_method='env://')
synchronize()
device = 'cuda'
train_set = COCODataset(args.path, 'train',
preset_transform(args, train=True))
valid_set = COCODataset(args.path, 'val',
preset_transform(args, train=False))
# backbone = vovnet39(pretrained=True)
# backbone = vovnet57(pretrained=True)
# backbone = resnet18(pretrained=True)
backbone = resnet50(pretrained=True)
#backbone = resnet101(pretrained=True)
model = FCOS(args, backbone)
model = model.to(device)
optimizer = optim.SGD(
model.parameters(),
lr=args.lr,
momentum=0.9,
if __name__ == '__main__':
args = get_args()
n_gpu = int(os.environ['WORLD_SIZE']) if 'WORLD_SIZE' in os.environ else 1
args.distributed = n_gpu > 1
if args.distributed:
torch.cuda.set_device(args.local_rank)
torch.distributed.init_process_group(backend='nccl',
init_method='env://')
synchronize()
device = 'cuda'
train_set = COCODataset(args.path, args.domain, 'train',
preset_transform(args, train=True))
valid_set = COCODataset(args.path, args.domain, 'val',
preset_transform(args, train=False))
target_train_set = COCODataset(args.path, args.target_domain, 'train',
preset_transform(args, train=True))
target_valid_set = COCODataset(args.path, args.target_domain, 'val',
preset_transform(args, train=False))
backbone = resnet50(pretrained=False, if_include_top=False)
model = FCOS(args, backbone)
model = model.to(device)
g_params = [
p for n, p in model.named_parameters()
# img = self.transforms(img)
#
# return img, my_annotation
#
# def __len__(self):
# return len(self.ids)
#
#
# def collate_fn(batch):
# return tuple(zip(*batch))
if __name__ == '__main__':
root = 'C://DeepLearningData/COCOdataset2017/images/val'
annotation = 'C://DeepLearningData/COCOdataset2017/annotations/instances_val2017.json'
coco_dset = COCODataset(root, annotation,
transforms.Compose([transforms.ToTensor()]), True)
data_loader = torch.utils.data.DataLoader(coco_dset,
batch_size=1,
shuffle=False,
num_workers=0,
collate_fn=collate_fn)
for i, [image, annotation] in enumerate(data_loader):
image = image[0].permute(1, 2, 0)
ann = annotation[0]
print(ann['label'])
print(ann['category'])
print(ann['mask'][250])
plt.subplot(1, 2, 1)
plt.imshow(image)
if args.distributed:
torch.cuda.set_device(args.local_rank)
torch.distributed.init_process_group(backend='gloo',
init_method='env://')
synchronize()
device = 'cuda'
valid_trans = transform.Compose([
transform.Resize(args.test_min_size, args.test_max_size),
transform.ToTensor(),
transform.Normalize(args.pixel_mean, args.pixel_std)
])
valid_set = COCODataset("/data/COCO_17/", 'val', valid_trans)
# backbone = vovnet39(pretrained=False)
# backbone = resnet18(pretrained=False)
backbone = resnet50(pretrained=False)
model = ATSS(args, backbone)
# load weight
model_file = "./training_dir/epoch-12.pt"
chkpt = torch.load(model_file, map_location='cpu') # load checkpoint
model.load_state_dict(chkpt['model'])
print('load weights from ' + model_file)
model = model.to(device)
if args.distributed:
def train(config: dict = None):
# fix the seed for reproduce results
SEED = config["SEED"]
torch.manual_seed(SEED)
torch.cuda.manual_seed(SEED)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
random.seed(SEED)
# parse config parameters
DATA_ROOT = config["DATA_ROOT"]
COCO_PATH = config["COCO_PATH"]
DATA_ROOT_VAL = config["DATA_ROOT_VAL"]
COCO_PATH_VAL = config["COCO_PATH_VAL"]
EXPERIMENT_NAME = config["EXPERIMENT_NAME"]
OPTIMIZER_NAME = config["OPTIMIZER_NAME"]
OPTIMIZER_WEIGHT_DECAY = config["OPTIMIZER_WEIGHT_DECAY"]
OPTIMIZER_MOMENTUM = config["OPTIMIZER_MOMENTUM"]
OPTIMIZER_BETAS = config["OPTIMIZER_BETAS"]
OPTIMIZER_EPS = config["OPTIMIZER_EPS"]
OPTIMIZER_AMSGRAD = config["OPTIMIZER_AMSGRAD"]
OPTIMIZER_ADABOUND_GAMMA = config["OPTIMIZER_ADABOUND_GAMMA"]
OPTIMIZER_ADABOUND_FINAL_LR = config["OPTIMIZER_ADABOUND_FINAL_LR"]
LEARNING_RATE = config["LEARNING_RATE"]
LEARNING_RATE_STEP_SIZE = config["LEARNING_RATE_STEP_SIZE"]
LEARNING_RATE_GAMMA = config["LEARNING_RATE_GAMMA"]
TRAINABLE_BACKBONE_LAYERS = config["TRAINABLE_BACKBONE_LAYERS"]
RPN_ANCHOR_SIZES = config["RPN_ANCHOR_SIZES"]
RPN_ANCHOR_ASPECT_RATIOS = config["RPN_ANCHOR_ASPECT_RATIOS"]
RPN_PRE_NMS_TOP_N_TRAIN = config["RPN_PRE_NMS_TOP_N_TRAIN"]
RPN_PRE_NMS_TOP_N_TEST = config["RPN_PRE_NMS_TOP_N_TEST"]
RPN_POST_NMS_TOP_N_TRAIN = config["RPN_POST_NMS_TOP_N_TRAIN"]
RPN_POST_NMS_TOP_N_TEST = config["RPN_POST_NMS_TOP_N_TEST"]
RPN_NMS_THRESH = config["RPN_NMS_THRESH"]
RPN_FG_IOU_THRESH = config["RPN_FG_IOU_THRESH"]
RPN_BG_IOU_THRESH = config["RPN_BG_IOU_THRESH"]
BOX_DETECTIONS_PER_IMAGE = config["BOX_DETECTIONS_PER_IMAGE"]
LOG_FREQ = config["LOG_FREQ"]
COCO_AP_TYPE = config["COCO_AP_TYPE"]
TRAIN_SPLIT_RATE = config["TRAIN_SPLIT_RATE"]
BATCH_SIZE = config["BATCH_SIZE"]
NUM_EPOCH = config["NUM_EPOCH"]
DEVICE = config["DEVICE"]
NUM_WORKERS = config["NUM_WORKERS"]
# init directories
directories = Directories(experiment_name=EXPERIMENT_NAME)
# copy config file to experiment dir
yaml_path = os.path.join(directories.experiment_dir, "config.yml")
save_yaml(config, yaml_path)
# init tensorboard summary writer
writer = SummaryWriter(directories.tensorboard_dir)
# set pytorch device
device = torch.device(DEVICE)
if "cuda" in DEVICE and not torch.cuda.is_available():
print("CUDA not available, switching to CPU")
device = torch.device("cpu")
# use our dataset and defined transformations
dataset = COCODataset(
DATA_ROOT, COCO_PATH, get_transforms(config=config, mode="train")
)
if COCO_PATH_VAL:
dataset_val = COCODataset(
DATA_ROOT_VAL, COCO_PATH_VAL, get_transforms(config=config, mode="val")
)
else:
dataset_val = COCODataset(
DATA_ROOT, COCO_PATH, get_transforms(config=config, mode="val")
)
# +1 for background class
num_classes = dataset.num_classes + 1
config["NUM_CLASSES"] = num_classes
# add category mappings to config, will be used at prediction
category_mapping = get_category_mapping_from_coco_file(COCO_PATH)
config["CATEGORY_MAPPING"] = category_mapping
# split the dataset in train and val set if val path is not defined
if not COCO_PATH_VAL:
indices = torch.randperm(len(dataset)).tolist()
num_train = int(len(indices) * TRAIN_SPLIT_RATE)
train_indices = indices[:num_train]
val_indices = indices[num_train:]
dataset = torch.utils.data.Subset(dataset, train_indices)
dataset_val = torch.utils.data.Subset(dataset_val, val_indices)
# define training and val data loaders
data_loader_train = torch.utils.data.DataLoader(
dataset,
batch_size=BATCH_SIZE,
shuffle=True,
num_workers=NUM_WORKERS,
collate_fn=core.utils.collate_fn,
)
data_loader_val = torch.utils.data.DataLoader(
dataset_val,
batch_size=1,
shuffle=False,
num_workers=NUM_WORKERS,
collate_fn=core.utils.collate_fn,
)
# get the model using our helper function
model = get_torchvision_maskrcnn(
num_classes=num_classes,
trainable_backbone_layers=TRAINABLE_BACKBONE_LAYERS,
anchor_sizes=RPN_ANCHOR_SIZES,
anchor_aspect_ratios=RPN_ANCHOR_ASPECT_RATIOS,
rpn_pre_nms_top_n_train=RPN_PRE_NMS_TOP_N_TRAIN,
rpn_pre_nms_top_n_test=RPN_PRE_NMS_TOP_N_TEST,
rpn_post_nms_top_n_train=RPN_POST_NMS_TOP_N_TRAIN,
rpn_post_nms_top_n_test=RPN_POST_NMS_TOP_N_TEST,
rpn_nms_thresh=RPN_NMS_THRESH,
rpn_fg_iou_thresh=RPN_FG_IOU_THRESH,
rpn_bg_iou_thresh=RPN_BG_IOU_THRESH,
box_detections_per_img=BOX_DETECTIONS_PER_IMAGE,
pretrained=True,
)
# move model to the right device
model.to(device)
# construct an optimizer
params = [p for p in model.parameters() if p.requires_grad]
optimizer_factory = OptimizerFactory(
learning_rate=LEARNING_RATE,
momentum=OPTIMIZER_MOMENTUM,
weight_decay=OPTIMIZER_WEIGHT_DECAY,
betas=OPTIMIZER_BETAS,
eps=OPTIMIZER_EPS,
amsgrad=OPTIMIZER_AMSGRAD,
adabound_gamma=OPTIMIZER_ADABOUND_GAMMA,
adabound_final_lr=OPTIMIZER_ADABOUND_FINAL_LR,
)
optimizer = optimizer_factory.get(params, OPTIMIZER_NAME)
# and a learning rate scheduler
lr_scheduler = torch.optim.lr_scheduler.StepLR(
optimizer, step_size=LEARNING_RATE_STEP_SIZE, gamma=LEARNING_RATE_GAMMA
)
# create coco index
print("Creating COCO index...")
coco_api_train = get_coco_api_from_dataset(data_loader_train.dataset)
coco_api_val = get_coco_api_from_dataset(data_loader_val.dataset)
# train it for NUM_EPOCH epochs
for epoch in range(NUM_EPOCH):
best_bbox_05095_ap = -1
# train for one epoch, printing every PRINT_FREQ iterations
train_one_epoch(
model=model,
optimizer=optimizer,
data_loader=data_loader_train,
coco_api=coco_api_train,
device=device,
epoch=epoch,
log_freq=LOG_FREQ,
coco_ap_type=COCO_AP_TYPE,
writer=writer,
)
# update the learning rate
lr_scheduler.step()
# get iteration number
num_images = len(data_loader_train.dataset)
iter_num = epoch * num_images
# evaluate on the val dataset
loss_lists, coco_evaluator = evaluate(
model=model,
data_loader=data_loader_val,
coco_api=coco_api_val,
device=device,
iter_num=iter_num,
coco_ap_type=COCO_AP_TYPE,
writer=writer,
)
# update best model if it has the best bbox 0.50:0.95 AP
bbox_05095_ap = coco_evaluator.coco_eval["bbox"].stats[0]
if bbox_05095_ap > best_bbox_05095_ap:
model_dict = {"state_dict": model.state_dict(), "config": config}
torch.save(model_dict, directories.best_weight_path)
best_bbox_05095_ap = bbox_05095_ap
# save final model
model_dict = {"state_dict": model.state_dict(), "config": config}
torch.save(model_dict, directories.last_weight_path)
args = get_args()
args.threshold = 0.2
n_gpu = int(os.environ['WORLD_SIZE']) if 'WORLD_SIZE' in os.environ else 1
args.distributed = n_gpu > 1
if args.distributed:
torch.cuda.set_device(args.local_rank)
torch.distributed.init_process_group(backend='gloo',
init_method='env://')
synchronize()
device = 'cuda'
valid_set = COCODataset("/data/COCO_17/", 'val',
preset_transform(args, train=False))
# backbone = vovnet39(pretrained=False)
backbone = resnet18(pretrained=False)
model = FCOS(args, backbone)
# load weight
model_file = "./training_dir/epoch-1.pt"
chkpt = torch.load(model_file, map_location='cpu') # load checkpoint
model.load_state_dict(chkpt['model'])
print('load weights from ' + model_file)
model = model.to(device)
if args.distributed:
model = nn.parallel.DistributedDataParallel(
def main():
config_path = "config.yaml"
with open(config_path, 'r') as fp:
config = yaml.safe_load(fp)
coco_train_imgs = os.path.join(config['path'], "images/train2017")
coco_train_annos_path = os.path.join(
config['path'], "annotations/person_keypoints_train2017.json")
coco_val_imgs = os.path.join(config['path'], "images/val2017")
coco_val_annos_path = os.path.join(
config['path'], "annotations/person_keypoints_val2017.json")
epochs = config['num_epochs']
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
train_dataset = COCODataset(coco_train_imgs,
coco_train_annos_path,
num_keypoints=17,
transform=get_transform(train=True))
val_dataset = COCODataset(coco_val_imgs,
coco_val_annos_path,
num_keypoints=17,
transform=get_transform(train=False))
train_dataloader = DataLoader(train_dataset,
batch_size=8,
shuffle=True,
num_workers=4)
val_dataloader = DataLoader(val_dataset,
batch_size=4,
shuffle=True,
num_workers=4)
model = PoseModel(num_kpts=17)
model.to(device)
model.train()
criterion = JointMSELoss()
optimizer = optim.SGD(model.parameters(),
lr=config['learning_rate'],
momentum=config['momentum'])
logs = []
for e in range(epochs):
running_loss = 0.0
for i, data in enumerate(train_dataloader):
imgs, _, heatmaps = data
imgs = imgs.to(device)
heatmaps = heatmaps.to(device)
optimizer.zero_grad()
outs = model(imgs)
loss = criterion(outs.float(), heatmaps.float())
loss.backward()
optimizer.step()
running_loss += loss.item()
if i % 1000 == 999:
print("Epoch {} | Iteration {} | Loss {:.4f}".format(
e + 1, i + 1, running_loss / 2000.))
logs.append("Epoch {} | Iteration {} | Loss {:.4f}".format(
e + 1, i + 1, running_loss / 2000.))
running_loss = 0.0
validate(model, val_dataloader, criterion, device)
model.train()
print("Save model")
torch.save(model.state_dict(), f'model_epochs_{e}.pth')
print("Test training finished.")
torch.save(model.state_dict(), f'model_epochs_{epochs}.pth')
with open("logs.txt", 'a') as fp:
for el in logs:
fp.write(el + "\n")
print("Logs saved")
import torch.nn as nn
import torch
from loss import YOLO_loss
from dataset import COCODataset, print_result
from saver import Saver
from network import YOLO
from utils import join
num_epochs = 10
learning_rate = 1e-6
data_root = r'E:/LocalData/coco/train2017'
save_root = r'C:\LocalData'
data_loader = torch.utils.data.DataLoader(dataset=COCODataset(root=data_root),
batch_size=20,
shuffle=True)
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
model = YOLO()
if torch.cuda.device_count() > 1:
print("Let's use", torch.cuda.device_count(), "GPUs!")
model = nn.DataParallel(model)
model.to(device)
criterion = YOLO_loss()
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)
yolo_saver = Saver(model,
join(save_root, r'yolo', r'save'),
"yolo",
max_to_keep=20)
loaded_index = yolo_saver.load()
device = 'cuda'
train_trans = transform.Compose([
transform.RandomResize(args.train_min_size_range, args.train_max_size),
transform.RandomHorizontalFlip(0.5),
transform.ToTensor(),
transform.Normalize(args.pixel_mean, args.pixel_std)
])
valid_trans = transform.Compose([
transform.Resize(args.test_min_size, args.test_max_size),
transform.ToTensor(),
transform.Normalize(args.pixel_mean, args.pixel_std)
])
train_set = COCODataset(args.path, 'train', train_trans)
valid_set = COCODataset(args.path, 'val', valid_trans)
# backbone = vovnet39(pretrained=True)
# backbone = vovnet57(pretrained=True)
# backbone = resnet18(pretrained=True)
backbone = resnet50(pretrained=True)
#backbone = resnet101(pretrained=True)
model = ATSS(args, backbone)
model = model.to(device)
optimizer = optim.SGD(
model.parameters(),
lr=args.lr,
momentum=0.9,
weight_decay=0.0001,
def main(args):
# setup
writer = SummaryWriter(log_dir='./logs/{}'.format(args.run_name))
dev = 'cpu' if args.gpu is None else 'cuda:{}'.format(args.gpu)
device = torch.device(dev)
with open(args.config_file, 'r') as f:
try:
config = yaml.safe_load(f)
except yaml.YAMLError as exc:
logging.error("Invalid YAML")
logging.error(exc)
# setup models
model = ED()
model = model.to(device)
# setup training optimizers
learning_rate = args.learning_rate
optimizer = optim.Adam(model.parameters(), lr=learning_rate)
scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer, 'min')
# setup colmap runner
dataset_dir = args.dataset_dir
if(args.dataset_type == 0):
coco_dataset_dir = path.join(dataset_dir, 'coco')
train_dataset = COCODataset(coco_dataset_dir, config, mode='train')
val_dataset = COCODataset(coco_dataset_dir, config, mode='validation')
else:
raise NotImplementedError
epoch_start = 0
epoch_end = args.epochs
total_step = 0
# load from checkpoint iff exists
latest_checkpoint_name = '{}-latest.ckpt'.format(args.run_name)
latest_checkpoint_path = path.join(args.checkpoint_dir, latest_checkpoint_name)
if((path.exists(latest_checkpoint_path)) and args.resume):
checkpoint = CheckPoint.load(latest_checkpoint_path, device)
model_weight = checkpoint['model']
optimizer_weight = checkpoint['optimizer']
scheduler_weight = checkpoint['scheduler']
epoch_start = checkpoint['epoch']
total_step = checkpoint['total_step']
model.load_state_dict(model_weight)
optimizer.load_state_dict(optimizer_weight)
scheduler.load_state_dict(scheduler_weight)
# setup dataset and actually run training
for epoch in range(epoch_start, epoch_end):
####################
# run training
model.train()
sample_indices = list(range(len(train_dataset)))
random.shuffle(sample_indices)
sample_it = tqdm(sample_indices, leave=False)
for data_index in sample_it:
data = train_dataset[data_index]
images, corrs, angles = data
images = images.to(device)
corrs = corrs.to(device)
angles = angles.to(device)
# compute loss and update
normals = model(images)
loss = surface_normal_loss(normals, corrs, angles)
# update weights
optimizer.zero_grad()
loss.backward()
optimizer.step()
writer.add_scalar('data/loss', loss, total_step)
sample_it.set_description('loss: {:.02f}'.format(loss))
if((total_step) % args.image_step == 0):
writer.add_images('images/normals', (normals + 1 / 2), total_step)
total_step += 1
# save every epoch
checkpoint_name = '{}-{}.ckpt'.format(args.run_name, epoch)
checkpoint_path = path.join(args.checkpoint_dir, checkpoint_name)
CheckPoint.save(checkpoint_path, model, optimizer, scheduler, total_step, epoch)
CheckPoint.save(latest_checkpoint_path, model, optimizer, scheduler, total_step, epoch)
writer.close()
r'''上下文管理测试代码块运行时间,需要
import time
from contextlib import contextmanager
'''
start = time.perf_counter()
try:
yield
finally:
end = time.perf_counter()
print('\033[1;34m{} : {}\033[0m'.format(label, end - start))
xml_dir = "/home/chiebotgpuhq/MyCode/dataset/Siam_detection/aqmzc"
json_path = "/home/chiebotgpuhq/Share/gpu-server/disk/disk1/coco_dataset/annotations/instances_val2017.json"
# a=XMLLikeDataset(xml_dir)
a = COCODataset(json_path)
# siam_g=GenerateSiamsesSample(a,"/home/chiebotgpuhq/MyCode/dataset/test")
# siam_g.test()
# with timeblock("too slow"):
b, f, m = a[1263]
# for b,f,m in a:
# print(type(b))
# cv2.imshow("b",b)
# cv2.imshow("f",f)
# cv2.imshow("m",m)
# cv2.waitKey(100)
# b,roi=seamlessclone(b,f,m)
# plt.imshow(b)
plt.imshow(m)
# plt.imshow(m)
#存图片
cv2.imwrite(os.path.join(save_dir, img_name), img)
cv2.imwrite(os.path.join(save_dir, img_o_name), img_o)
save_xml(save_dir, idx, img, box)
def test(self):
bg, fg, mask = self.dataset[100]
self.deal_one_sample(bg, fg, mask, 100)
def do_task(self, max_time: int = None, work_num: int = 8):
if max_time is None:
max_time = len(self.dataset)
'''
XXX:测试代码
'''
for idx in tqdm(range(4589, max_time)):
self.deal_one_sample(*self.dataset[idx], idx)
# with futures.ThreadPoolExecutor() as exec:
# task_list=(exec.submit(self.deal_one_sample,*self.dataset[idx],idx) \
# for idx in range(max_time))
# for task in tqdm(futures.as_completed(task_list),total=max_time):
# pass
if __name__ == "__main__":
json_path = "/home/chiebotgpuhq/Share/gpu-server/disk/disk1/coco_dataset/annotations/instances_val2017.json"
save_dir = "/home/chiebotgpuhq/Share/gpu-server/disk/disk2/cocosiam_dataset/val"
dataset = COCODataset(json_path, sampler='normal')
test_cls = GenerateSiamsesSample(dataset, save_dir)
test_cls.do_task()
def main(args):
print(args)
traintrans = transforms.Compose([
transforms.Resize((args.image_size, args.image_size)),
transforms.ToTensor()
])
traindset = COCODataset(args.root_dir,
args.ann_dir,
mode="train",
transform=traintrans)
trainloader = DataLoader(traindset,
batch_size=args.batch_size,
shuffle=True,
num_workers=args.ncpu,
collate_fn=utils.collate_fn,
drop_last=True)
"""
create dictionary from args.captionfile
"""
vocab = Dictionary()
utils.create_vocfile(traindset, args.captionfile)
before = time.time()
vocab.create_vocab(args.captionfile)
print('took {}s for vocab'.format(time.time() - before), flush=True)
"""
models for training
"""
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
yolo = YOLOv3()
capmodel = Captioning(vocab_size=len(vocab))
if args.multi_gpu:
print('Using {} GPUs...'.format(torch.cuda.device_count()))
if torch.cuda.device_count() > 1:
capmodel = nn.DataParallel(capmodel)
yolo = nn.DataParallel(yolo)
if args.use_gpu:
capmodel = capmodel.to(device)
yolo = yolo.to(device)
mseloss = nn.MSELoss()
bceloss = nn.BCELoss()
celoss = nn.CrossEntropyLoss()
anchors = torch.tensor([[10, 13], [16, 30], [33, 23], [30, 61], [62, 45],
[59, 119], [116, 90], [156, 198], [373, 326]],
dtype=torch.float).to(device)
scaled_anchors = utils.scale_anchors(args.image_size, args.grid_size,
anchors)
print('begin training')
for ep in range(args.epochs):
for it, sample in enumerate(trainloader):
# sample contains 'image', 'captions', 'bboxinfo'
im_ten = sample['image'].to(device)
info = sample['bboxinfo']
"""
for objects in info:
bounding_boxes = []
classes = []
for bbinfo in objects:
bounding_boxes.append(bbinfo['bbox'])
classes.append(bbinfo['obj_id'])
bbs = torch.tensor(bounding_boxes)
cls = torch.tensor(classes)
print(bbs.size())
print(cls.size())
im_ten = im_ten.to(device)
"""
x, y, w, h, pred_conf, pred_cls = yolo(im_ten)
pred_x, pred_y, pred_w, pred_h = utils.offset_boxes(
x, y, w, h, device, args.image_size, args.grid_size,
scaled_anchors)
break
break
print('done training')
""" # size check of captioning model
