weightfile1 = 'reid.weights'
cfgfile2 = 'reid_nbn.cfg'
weightfile2 = 'reid_nbn.weights'
cfgfile3 = 'reid_nbn.prototxt'
weightfile3 = 'reid_nbn.caffemodel'
m1 = Darknet(cfgfile1)
m1.load_weights(weightfile1)
m1.eval()
m2 = Darknet(cfgfile2)
m2.load_weights(weightfile2)
m2.eval()
m3 = CaffeNet(cfgfile3)
m3.load_weights(weightfile3)
m3.eval()
img = torch.rand(8, 3, 128, 64)
img = Variable(img)
output1 = m1(img).clone()
output2 = m2(img).clone()
output3 = m3(img).clone()
print('----- output1 ------------------')
print(output1.data.storage()[0:100])
print('----- output2 ------------------')
print(output2.data.storage()[0:100])
print('----- output3 ------------------')
print(output3.data.storage()[0:100])
net.cuda()
optimizer = optim.SGD(net.parameters(), lr=base_lr, momentum=momentum, weight_decay=weight_decay)
if args.weights:
state = torch.load(args.weights)
start_epoch = state['batch']+1
model.load_state_dict(state['state_dict'])
optimizer.load_state_dict(state['optimizer'])
print('loaded state %s' % (args.weights))
net.train()
buf = Variable(torch.zeros(len(device_ids)).cuda())
for batch in range(max_iter):
if (batch+1) % test_interval == 0:
net.eval()
average_accuracy = 0.0
average_loss = 0.0
for i in range(test_iter):
loss, accuracy = net(buf)
average_accuracy += accuracy.data.mean()
average_loss += loss.data.mean()
average_accuracy /= test_iter
average_loss /= test_iter
print('[%d] test loss: %f\ttest accuracy: %f' % (batch+1, average_loss, average_accuracy))
net.train()
else:
optimizer.zero_grad()
loss = net(buf).mean()
loss.backward()
optimizer.step()
parser.add_argument('--gpu', help='gpu ids e.g "0,1,2,3"')
parser.add_argument('--solver', help='the solver prototxt')
parser.add_argument('--model', help='the network definition prototxt')
parser.add_argument('--snapshot', help='the snapshot solver state to resume training')
parser.add_argument('--weights', help='the pretrained weight')
args = parser.parse_args()
if args.mode == 'time':
protofile = args.model
net_info = parse_prototxt(protofile)
model = CaffeNet(protofile)
batch_size = 64
model.print_network()
model.eval()
niters = 50
total = 0
for i in range(niters):
x = torch.rand(batch_size, 1, 28, 28)
x = Variable(x)
t0 = time.time()
output = model(x)
t1 = time.time()
print('iteration %d: %fs' %(i, t1-t0))
total = total + (t1-t0)
print('total time %fs, %fs per batch, %fs per sample' % (total, total/niters, total/(niters*batch_size)))
elif args.mode == 'train':
solver = parse_prototxt(args.solver)