def eval_define(self,weights_path):
n_classes = self.n_classes
net = ShelfNet(n_classes=n_classes)
net.load_state_dict(torch.load(weights_path))
net.cuda()
net.eval()
self.evaluator = MscEval(net, dataloader=None, scales=[1.0],flip=False)
def evaluate(respth='./res',
dspth='/data2/.encoding/data/cityscapes',
checkpoint=None):
## logger
logger = logging.getLogger()
## model
logger.info('\n')
logger.info('====' * 20)
logger.info('evaluating the model ...\n')
logger.info('setup and restore model')
n_classes = 19
net = ShelfNet(n_classes=n_classes)
if checkpoint is None:
save_pth = osp.join(respth, 'model_final.pth')
else:
save_pth = checkpoint
net.load_state_dict(torch.load(save_pth))
net.cuda()
net.eval()
## dataset
batchsize = 5
n_workers = 2
dsval = CityScapes(dspth, mode='val')
dl = DataLoader(dsval,
batch_size=batchsize,
shuffle=False,
num_workers=n_workers,
drop_last=False)
## evaluator
logger.info('compute the mIOU')
evaluator = MscEval(net, dl, scales=[1.0], flip=False)
## eval
mIOU = evaluator.evaluate()
logger.info('mIOU is: {:.6f}'.format(mIOU))
def test():
model = ShelfNet(n_classes=19)
# official implementation
#model = BiSeNet(19,criterion=None,ohem_criterion=None,is_training=False)
print(model)
# count parameter number
pytorch_total_params = sum(p.numel() for p in model.parameters())
print("Total number of parameters: %.3fM" % (pytorch_total_params / 1e6))
model = model.cuda()
model.eval()
run_time = list()
for i in range(0, 100):
input = torch.randn(1, 3, 1024, 2048).cuda()
# ensure that context initialization and normal_() operations
# finish before you start measuring time
torch.cuda.synchronize()
torch.cuda.synchronize()
start = time.perf_counter()
with torch.no_grad():
#output = model(input)#, aucx=False)
output = model(input, aux=False)
torch.cuda.synchronize() # wait for mm to finish
end = time.perf_counter()
print(end - start)
run_time.append(end - start)
run_time.pop(0)
print('Mean running time is ', np.mean(run_time))