'''filein = r'/home/liming/code/get_mat/ballet_106_1_0_with.png'
filein2 = r'/home/liming/code/get_mat/ballet_106_1_0_with_mask.png'
width = 100
height = 32
type = 'png'
ResizeImage(filein, 'input.png', width, height, type)
ResizeImage(filein2, 'mask.png', width, height, type)'''
# 最终将图片保存至哪里
result_path = '../net_result3d_64bs'
if not os.path.exists(result_path):
os.mkdir(result_path)
criterion = nn.MSELoss()
model = resNet.resnext50(num_classes=9600)
model = model.cuda()
model = nn.DataParallel(model)
image = Image.open('input.png').convert('RGB')
image = loader(image).unsqueeze(0)
input = image.to(t.float)
image = Image.open('mask.png').convert('RGB')
image = loader(image).unsqueeze(0)
mask = image.to(t.float)
# mask = mask[0][2]
# mask = mask[:,2,:,:]
# print(mask.size())
input = input.cuda()
nc = 4
criterion = nn.MSELoss()
# custom weights initialization called on crnn
def weights_init(m):
classname = m.__class__.__name__
if classname.find('Conv') != -1:
m.weight.data.normal_(0.0, 0.02)
elif classname.find('BatchNorm') != -1:
m.weight.data.normal_(1.0, 0.02)
m.bias.data.fill_(0)
model_mask = resNet.resnext50(num_classes=3200)
model_bb = crnn.CRNN(opt.imgH, nc, nclass, opt.nh)
model_bb.apply(weights_init)
if opt.pretrained != '':
print('loading pretrained model from %s' % opt.pretrained)
model_bb.load_state_dict(torch.load(opt.pretrained))
print(model_bb)
image_rgb = torch.FloatTensor(opt.batchSize, 3, opt.imgH, opt.imgW)
image_2channel = torch.FloatTensor(opt.batchSize, 2, opt.imgH, opt.imgW)
bb = torch.FloatTensor(opt.batchSize, 26, 8)
'''text = torch.IntTensor(opt.batchSize * 5)
length = torch.IntTensor(opt.batchSize)'''
if opt.cuda:
model_mask.cuda()