transforms.Resize((299, 299)),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])
import glob
pictures = glob.glob('/root/picture/*.png')
pictures = [Image.open(picture).convert('RGB') for picture in pictures]
pictures = [transform(face) for face in pictures]
real_faces = [transform(face) for face in real_faces]
fake_faces = [transform(face) for face in fake_faces]
lxr_net = lxrEfficientNet.from_name('efficientnet-b4',
override_params={'num_classes': 2})
lxr_net = MyModel(lxr_net).cuda().eval()
lxr_net.load_state_dict(torch.load('/root/data/wzy/new_mtcnn_lxr.pth'))
online_perforance = {}
with torch.no_grad():
for num_fake, num_real, num_picture in [[12, 0, 0], [9, 0, 3], [6, 0, 6],
[3, 0, 9]]:
fake_linespace_num = np.linspace(start=0,
stop=len(fake_faces) - 1,
num=num_fake,
dtype=int)
real_linespace_num = np.linspace(start=0,
stop=len(real_faces) - 1,
num=num_real,
dtype=int)
picture_linespace_num = np.linspace(start=0,
import torch
from lxrlocker.kagglefile.kaggle_lxr import EfficientNet, MyModel
net = EfficientNet.from_name('efficientnet-b4', override_params={'num_classes': 2})
net = MyModel(net)
weight = {}
for key, value in torch.load('/root/data/wzy/checkpoint/efficient-all-frame-new/6.pth').items():
if '_fc' in key:
fc_key = key.replace('_fc', 'last_linear')
weight[fc_key] = value
key = 'backbone.' + key
print(key)
weight[key] = value
# net.load_state_dict(weight)
# torch.save(net.state_dict(), "/root/data/wzy/lxrlocker/efficient-Acc0.875-All-Frame-45part-epoch-6.pth")
import cv2
from tqdm import tqdm
import os
import random
from model.efficientnet.model import EfficientNet
from torch.nn import functional as F
random.seed(0)
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
from lxrlocker.kagglefile.kaggle_lxr import MyModel, EfficientNet as lxrEfficientNet
wzy_net = EfficientNet.from_name('efficientnet-b4', override_params={'num_classes': 2}).cuda().eval()
wzy_net.load_state_dict(torch.load('/root/data/wzy/checkpoint/efficient-video/14.pth'))
lxr_net = lxrEfficientNet.from_name('efficientnet-b4', override_params={'num_classes': 2})
lxr_net = MyModel(lxr_net).cuda().eval()
lxr_net.load_state_dict(torch.load('/root/data/wzy/statistic/new_50partpai_tracking2_resize.pth'))
from facenet_pytorch import MTCNN
import numpy as np
class FastMTCNN(object):
def __init__(self, *args, **kwargs):
self.mtcnn = MTCNN(*args, **kwargs)
def __call__(self, frame):
boxes, scores = self.mtcnn.detect([frame])
boxes, scores = boxes[0], scores[0]
new_boxes = []
for score_index, face_score in enumerate(scores):
print('==> Preparing data..')
BATCH_SIZE = 32
EPOCH = 20
INTERVAL = 200
model_name = 'efficient-all-frame-new'
save_path = join("/root/data/wzy/checkpoint", model_name)
train_dataloader, test_dataloader = get_dataloader(BATCH_SIZE)
print('==> Building model..')
net = EfficientNet.from_name('efficientnet-b4',
override_params={
'num_classes': 2
}).eval()
net = MyModel(net).cuda()
net.load_state_dict(
torch.load(
'/root/data/wzy/lxrlocker/scale2models_lxr/45partmtcnnenhance.pth'))
wzy_net = wzyEfficient.from_name('efficientnet-b4',
override_params={
'num_classes': 2
}).cuda()
# model_dict = wzy_net.state_dict()
#
# weight = {}
# for key, value in torch.load('/root/data/wzy/lxrlocker/scale2models_lxr/new_45partpai_tracking2_resize.pth').items():
# key = key.replace('backbone.', '')
# if '_fc' in key:
# continue
# key = key.replace('last_linear', '_fc')