def save_network(network, epoch_label):
save_filename = 'net_%s.pth' % epoch_label
save_path = os.path.join('./checks/mpn_no_ca/1501/1/',save_filename)
torch.save(network.cpu().state_dict(), save_path)
if torch.cuda.is_available:
network.to(device)
#########################################################
# define student
TSModel = TS(num_classes=751, num_stripes=6)
TSModel = TSModel.to(device)
##########################################################
# set the criterion
triplet_selector_S = Hard(opt.margin1)
criterion_tri_S = OnlineTripletLoss(opt.margin1, triplet_selector_S)
criterion_part_S = nn.CrossEntropyLoss()
criterion_part_T = nn.CrossEntropyLoss()
criterion_cosine = torch.nn.CosineSimilarity(dim=0, eps=1e-6)
param_groups = [{'params': TSModel.parameters(), 'lr': 0.01}]
optimizer_ft = optim.SGD(
param_groups,
momentum=0.9, weight_decay=5e-4, nesterov=True)
# rule for learning rate
exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft, step_size=20, gamma=0.1)
save_path = os.path.join(opt.outf, save_filename)
torch.save(network.cpu().state_dict(), save_path)
if torch.cuda.is_available:
network.cuda()
# setting of train phase
model = PCB(class_num=4768)
# load pretrained para without classifier
if use_gpu:
model = model.cuda()
# set the criterion
triplet_selector = SemihardNegativeTripletSelector(opt.margin)
criterion_tri = OnlineTripletLoss(opt.margin, triplet_selector)
criterion_part = nn.CrossEntropyLoss()
# criterion_part=CrossEntropyLabelSmooth(4768)
criterion_center = CenterLoss(4768)
criterion_focal = FocalLoss(gamma=2)
# updating rule for parameter
ignored_params = list(map(id, model.model.fc.parameters()))
ignored_params += (
list(map(id, model.classifier0.parameters())) +
list(map(id, model.classifier1.parameters())) +
list(map(id, model.classifier2.parameters())) +
list(map(id, model.classifier3.parameters())) +
list(map(id, model.classifier4.parameters())) +
list(map(id, model.classifier5.parameters())) +